Internet of Things (IoT) devices generate large volumes of time series data that are often volatile and complex, making probabilistic time series forecasting (TSF) essential for modeling the distribution of future outcomes. Recently, diffusion-based TSF methods have gained attention for their ability to learn complex distributions. However, they typically apply the diffusion process directly in the time domain, which may struggle to capture complex temporal dependencies, thus limiting the full potential of the diffusion process. Besides, they obtain probabilistic forecasts by sampling multiple plausible outcomes from the learned distribution, which is time-consuming and less effective. To solve these problems, we propose Invertible Latent Diffusion for probabilistic Time Series Forecasting (IL-DiffTSF), a novel approach based on a latent diffusion model. Specifically, we design an invertible latent projection between time series and latent space, where a conditional diffusion process is applied. This design ensures bidirectional consistency and minimal information loss, enabling more accurate TSF. Moreover, instead of sampling-based probabilistic forecasting, IL-DiffTSF represents uncertainties by directly learning a mapping from latent representations to prediction errors, achieving faster and more reliable uncertainty estimates. Experiments on univariate and multivariate benchmarks validate the efficiency and effectiveness of IL-DiffTSF. The code for this project is available at https://github.com/vanerkz/IL-DiffTSF.

Box-free model watermarking is an emerging technique to safeguard the intellectual property of deep learning models, particularly those for low-level image processing tasks. Existing works have verified and improved its effectiveness in several aspects. However, in this paper, we systematically investigate the vulnerability and demonstrate that box-free model watermarking is prone to removal attacks, even under the real-world threat model such that the protected model and the watermark extractor are in black boxes. Under this setting, we carry out three studies. 1) We develop an extractor-gradient-guided (EGG) remover and show its effectiveness when the extractor uses ReLU activation only. 2) More generally, for an unknown extractor, we leverage adversarial attacks and design the EGG remover based on the estimated gradients. 3) Under the most stringent condition that the extractor is inaccessible, we design a transferable remover based on a set of private proxy models. In all cases, the proposed removers can successfully remove embedded watermarks while preserving the quality of the processed images, and we also demonstrate that the EGG remover can even replace the watermarks. Extensive experimental results verify the effectiveness and generalizability of the proposed attacks, revealing the vulnerabilities of the existing box-free methods and calling for further research.

Generative adversarial networks (GANs) are a set of powerful generative models, among which CycleGAN, featuring the unique cycle-consistency loss, has gained special popularity. However, this unique structure and the cycle-consistency loss make watermarking CycleGAN particularly challenging, rendering existing deep neural network (DNN) watermarking methods, whether model-agnostic or GAN-specific, inapplicable. Meanwhile, existing DNN watermarking methods are intrusive in nature, requiring direct or indirect modification of model parameters for watermark embedding, which raises fidelity concerns. To solve the above problems, we propose the first nonintrusive and robust watermarking method for CycleGAN. We empirically show that without modifying the CycleGAN model, a user-defined watermark image can still be extracted from model outputs using a dedicated watermark decoder. Extensive experimental results verify that while achieving the so-called absolute fidelity, the proposed method is robust to various attacks, from image post-processing to model stealing.

The language modeling paradigm for scene text recognition (STR) has demonstrated impressive universal capabilities across extensive STR scenarios. However, existing methods still encounter challenges in effectively handling text images with irregular shapes and diverse appearances (e.g., curve, artistic, multi-oriented) due to the absence of contextual information during initial decoding. In this work, inspired by the principle of ‘forest before trees’ in human visual perception, we introduce NASTR, a non-autoregressive scene text recognizer capable of endowing global-aware for the attentional decoder. Specifically, we design a global-to-local attention procedure, simulating the mechanism of globally holistic visual signal processing preceding locally detailed response in the human brain visual system. This is achieved by leveraging the global image information queries to condition the generation of glimpse vectors at each decoding time step. This procedure empowers the NASTR model to achieve on-par performance with its state-of-the-art autoregressive counterparts, while operating in a fully parallel manner. Moreover, we propose multiple optional and flexible encoding constraint components to alleviate the representation quality degradation issue caused by the global image information queries in handling STR tasks with multilingual and in multi-domains. These components constrain the global image features from the perspective of global structural, global semantic, and linguistic knowledge. Extensive experimental results demonstrate that NASTR consistently outperforms existing methods on both Chinese and English STR benchmarks. Our source code, trained models, and logs are available at https://github.com/ML-HDU/NASTR.

Multi-modal image fusion aims to integrate complementary cues from different modalities into a single image, facilitating downstream tasks such as object detection. However, input image pairs are often misaligned due to rigid or non-rigid deformations in practical scenarios. Such misregistration introduces structural distortions and visual artifacts, reducing the reliability of the fused results and limiting their effectiveness for downstream vision applications. While existing methods demonstrate satisfactory results under specific deformation scenarios, they exhibit limited generalization to diverse and severe misregistrations. To this end, this study proposes a unified multiple-in-one flow Mamba framework for registering various image deformations and generating high-quality fused results. Specifically, a hierarchical flow Mamba is designed to model rigid and non-rigid flow fields and enhance adaptability to complex deformations by progressively refining misaligned features. To better distinguish between rigid and non-rigid types, a flow field classifier predicts rigid/non-rigid categories and provides prompts for high-level feature modulation. Furthermore, a flow-guided fusion Mamba module is developed to aggregate aligned multi-level modality features and generate fused images, while an iterative training strategy enables collaborative optimization by using fusion outputs to refine flow estimation. Experiments across three representative modality tasks demonstrate that the proposed method delivers superior fusion performance while maintaining applicability to object detection. The code will be available at: https://github.com/BOYang-pro/MOFM.

Network traffic classification is crucial for both network security and management. Despite advances in deep learning-based multi-task traffic classification, existing models often struggle to jointly handle multiple tasks while providing interpretable insights. In multi-task scenarios, different tasks rely on distinct regions of the traffic sequence, motivating the use of dynamic and interpretable attention mechanisms. To this end, we propose Dynamic Quantized Self-Attention (DQSA), a unified framework specifically designed for multi-task network traffic classification. At its core, the Task Gated Attention Router (TGAR) dynamically associates attention heads with different tasks, enabling adaptive focus on task-specific patterns. This mechanism provides interpretable attention scores, which help analyze misclassifications and guide further model refinement. To improve efficiency and handle diverse network traffic features, we introduce the Soft Quantized Self-Attention Head (SQ-SAH) to reduce computational complexity and extend the Rotary Position Embedding (RoPE) to accommodate these features. Extensive experiments on ISCX VPN-NonVPN and DCI-LTE datasets demonstrate that DQSA consistently outperforms state-of-the-art baselines, achieving 92.85% accuracy on the encapsulation-level task of ISCX VPN-NonVPN and 93.17% accuracy on the application-level task of DCI-LTE, surpassing the strongest existing methods by up to 2.65%, while providing interpretable task-specific attention for efficient multi-task network traffic classification.

In light of their capability to capture structural information while reducing computing complexity, anchor graph-based multi-view clustering (AGMC) methods have attracted considerable attention in large-scale clustering problems. Nevertheless, existing AGMC methods still face the following two issues: 1) They directly embedded diverse anchor graphs into a consensus anchor graph (CAG), and hence ignore redundant information and numerous noises contained in these anchor graphs, leading to a decrease in clustering effectiveness; 2) They drop effectiveness and efficiency due to independent post-processing to acquire clustering indicators. To overcome the aforementioned issues, we deliver a novel one-step multi-view clustering method with adaptive low-rank anchor-graph learning (OMCAL). To construct a high-quality CAG, OMCAL provides a nuclear norm-based adaptive CAG learning model against information redundancy and noise interference. Then, to boost clustering effectiveness and efficiency substantially, we incorporate category indicator acquisition and CAG learning into a unified framework. Numerous studies conducted on ordinary and large-scale datasets indicate that OMCAL outperforms existing state-of-the-art methods in terms of clustering effectiveness and efficiency.

Time series anomaly detection tools (TSAD) are widely applicable across industries, such as monitoring time series data of water pipeline pressure, network traffic activities, and hardware telemetry. The primary objective is to identify anomalous segments and alert users to potential issues before any consequences. A closely related tool is time series forecasting, and some practitioners leverage it for anomaly detection. As the main objective of a forecasting model is to minimize errors, it tends to over-fit the time series, making it challenging to distinguish whether the forecasting errors occur due to model limitations or true anomalous segments. This paper introduces a method called the posterior anomaly scoring criterion, which uses deep learning time series forecasting models to estimate forecast uncertainties for TSAD. We propose replacing the forecasting model’s output layer to estimate forecast distributions and compute the probability of the posterior distribution to attain anomaly scores. These scores are processed through an automated threshold criterion to classify the anomalous segments. The experiments have demonstrated our model performs the best in four out of five datasets across seven benchmark models.

Multivariate Time Series (MTS) anomaly detection is challenging in distinguishing anomalous data from normal data in high-dimensional, complex distributions. Even some deep learning methods still have difficulties capturing intricate MTS patterns. Recent advancements in Diffusion Models (DM) for sample generation have inspired us to explore their potential in MTS anomaly detection. In this paper, TWavefussion, an unsupervised diffusion model for MTS anomaly detection combining wavelet-based diffusion model and transformer autoencoder, is proposed. The wavelet-based diffusion model captures fine-grained local features in the high-frequency components of latent features and helps fuse both local and global MTS features better. Comparative experiments show TWavefussion achieves leading performance on three of four datasets.

The advancement of Integrated Sensing and Communication (ISAC) has facilitated the adoption of target tracking technologies, yet most existing research focuses on waveform design rather than specific tracking algorithms within ISAC systems. To address this gap, this paper proposes an effective target tracking method based on Kalman filter (KF) algorithms within the ISAC framework. The method filters noisy radar measurements, such as bistatic range (BR), bistatic range rate (BRR), and direction of arrival (DOA), to enhance tracking accuracy. Performance is assessed by using trajectory maps and root-meansquare error (RMSE) curves, demonstrating that the proposed method significantly improves estimation accuracy while increasing robustness to measurement noise, especially in DOA. Compared to our previously proposed geometric target localization method, this approach delivers superior tracking performance, making it highly suitable for real-time applications in ISAC systems.

MUX-based logic locking (LL) is a hardware security technique that inserts multiplexers (MUX) into circuits to secure them against unauthorized use and reverse engineering by protecting original circuit pathways. Nevertheless, MUX-based LL is vulnerable to Oracle-Guided (OG) and Oracle-Less (OL) attacks. While OG methods, such as the SAT attack, are infeasible for large-scale designs, OL attacks, like those based on machine learning (ML), can exploit structural leakage in locked circuits to recover original pathways. This study introduces N-MUX, an innovative MUX-based LL approach designed to resist state-of-the-art (SOTA) ML attacks. N-MUX effectively reduces structural leakage by identifying maximal overlap structures in the original circuit to configure the MUX logic. Additionally, N-MUX ensures high efficiency by selecting the false input from the direct neighbourhood of the true input. Experimental results on ISCAS’85 and ITC’99 benchmarks demonstrate that N-MUX is the most secure and reliable LL technique against SOTA ML-based attacks, achieving an 81% reduction in attack accuracy compared to existing MUX-based LL methods and delivering up to 480× greater efficiency.

In this letter, we propose the Bi-nuclear Schatten-p norm minimization (BSNM) for coprime array to achieve the efficient direction of arrival (DOA) estimation. Specifically, BSNM factorizes a large Hermitian Toeplitz covariance matrix as the product of two small matrices, which not only improves the computational efficiency but also captures the low-rank property of the interpolated covariance matrix. This BSNM problem of the Hermitian Toeplitz matrix is then solved by a parallel alternating optimization algorithm. Finally, the recovered covariance matrix is applied to estimate the DOA via the MUSIC algorithm. Numerical simulation illustrates the superiority of the proposed BSNM method over state-of-the-art methods in terms of computational efficiency.

Hyperspectral target detection (HTD) is an approach to identify targets of interest in a scene by utilizing prior target spectra. Existing deep learning-based HTD methods usually need to generate a large number of samples for network training, and these generated samples often suffer from distortion. In addition, most of the methods can only be applied to a single scene. To address these issues, this study proposes a self-supervised meta-transfer learning (SelfMTL) method to improve the generalization ability and adaptability of the model through contrastive representation. First, labeled source data, which contains rich feature information, is utilized to train the global-local spectral contrastive learning (GLSL) module by randomly constructing positive and negative pairs from different land covers for the classification task, aiming to effectively discriminate the similarities and differences between spectra. Then, a small sample (only one target-background pair) fine-tuning is utilized to transfer the pretrained GLSL to different target detection (TD) tasks. Finally, a novel adaptive spatial-spectral enhancement (ASSE) module is proposed, which takes into account the joint learning constraints of spatial and spectral information to obtain the final detection result map. The experimental results on four real hyperspectral images (HSIs) datasets verify the superiority of SelfMTL in comparison to many classical and SOTA HTD methods. The codes are available athttps://github.com/ShissHAN/SelfMTL.

The use of learning prompts to adapt pretrained vision-language models (VLMs) for downstream tasks has gained significant attention due to its potential to reduce training costs compared to model fine-tuning through few-shot learning. Most existing methods rely on a universal prompt for all classes, as it generally delivers consistent performance across various datasets. However, a universal prompt cannot capture class-specific discriminative information. To overcome this limitation, we propose class-specific prompt learning (CPL). CPL represents the context of a prompt using two components: a base vector shared among all classes and a class-specific vector designed for individual classes. This method combines the generalization ability of the base context with the adaptability of the class-specific context. Furthermore, we introduce contrastive CPL, which enhances the ability of the prompt to capture discriminative features unique to each class. Also, we adopt the self-consistency loss to regularize the base context, enhancing its generalization ability. As a result, CPL effectively learns tailored prompts for each class. Extensive experiments demonstrate that CPL achieves superior performance over existing methods in both base-class classification and new class generalization.

Due to its high computational complexity, graph-based methods have limited applicability in large-scale multiview clustering tasks. To address this issue, many accelerated algorithms, especially anchor graph-based methods and indicator learning-based methods, have been developed and made a great success. Nevertheless, since the restrictions of the optimization strategy, these accelerated methods still need to approximate the discrete graph-cutting problem to a continuous spectral embedding problem and utilize different discretization strategies to obtain discrete sample categories. To avoid the loss of effectiveness and efficiency caused by the approximation and discretization, we establish a discrete fast multiview anchor graph clustering (FMAGC) model that first constructs an anchor graph of each view and then generates a discrete cluster indicator matrix by solving the discrete multiview graph-cutting problem directly. Since the gradient descent-based method makes it hard to solve this discrete model, we propose a fast coordinate descent-based optimization strategy with linear complexity to solve it without approximating it as a continuous one. Extensive experiments on widely used normal and large-scale multiview datasets show that FMAGC can improve clustering effectiveness and efficiency compared to other state-of-the-art baselines.

Class-incremental learning (CIL) under an exemplar-free constraint has presented a significant challenge. Existing methods adhering to this constraint are prone to catastrophic forgetting, far more so than replay-based techniques that retain access to past samples. In this paper, to solve the exemplar-free CIL problem, we propose a Dual-Stream Analytic Learning (DS-AL) approach. The DS-AL contains a main stream offering an analytical (i.e., closed-form) linear solution, and a compensation stream improving the inherent under-fitting limitation due to adopting linear mapping. The main stream redefines the CIL problem into a Concatenated Recursive Least Squares (C-RLS) task, allowing an equivalence between the CIL and its joint-learning counterpart. The compensation stream is governed by a Dual-Activation Compensation (DAC) module. This module re-activates the embedding with a different activation function from the main stream one, and seeks fitting compensation by projecting the embedding to the null space of the main stream's linear mapping. Empirical results demonstrate that the DS-AL, despite being an exemplar-free technique, delivers performance comparable with or better than that of replay-based methods across various datasets, including CIFAR-100, ImageNet-100 and ImageNet-Full. Additionally, the C-RLS' equivalent property allows the DS-AL to execute CIL in a phase-invariant manner. This is evidenced by a never-before-seen 500-phase CIL ImageNet task, which performs on a level identical to a 5-phase one. Our codes are available at https://github.com/ZHUANGHP/Analytic-continual-learning.

In this paper, we propose a novel pre-processing technique known as the Connectivity Matrix (CM). Building upon the foundation of the CM, we present an effective Second-Order Side-Channel Attack, called Connectivity Matrix Attack (CMA). Our work aims to efficiently counter hardware devices fortified with Masking countermeasures, and it contributes in three significant ways. First, the proposed CM has lower data complexity, as it is constant regarding the number of measurements. Second, we propose the decomposition of the CMs and the utilization of their eigenvalues as feature vectors in CMA. This approach effectively removes noisy components from the CMs and reduces their dimensions. Third, the proposed CMA employs the selected eigenvalues to establish a frequency distribution, followed by a chi-square test. This approach allows CMA to expose both the linear and non-linear leakages present in CMs. The proposed CMA is validated on the public dataset ASCAD and can reveal all the masked bytes successfully. Notably, the concept of the Connectivity Matrix extends beyond the confines of a correlation matrix used in this paper, opening the door to a promising avenue for future research.

COVID-19 exhibits diverse transmission routes and an extended incubation period, facilitating rapid dispersion across broad geographical regions. Therefore, the process of conducting COVID-19 detection holds utmost significance. Inspired by the idea of spectral signal-to-map, we introduce the Competitive Adaptive Reweighted Sampling-Principal Component Analysis (CARS-PCA) feature extractor in combination with a Convolutional Neural Network-Support Vector Machine (CNN-SVM) model, which aims to classify the infrared spectra obtained from COVID-19 pharyngeal swab samples. By conducting an experimental analysis on a series of infrared spectral signals obtained from positive and negative pharyngeal swabs, the experimental findings, on the one hand, indicate that the CARS-PCA algorithm is better suited for feature selection and reducing the dimensionality of the spectral signal data. On the other hand, they also illustrate the superior performance of the proposed CNN-SVM model in comparison to conventionally employed spectral signal recognition algorithms, achieving an accuracy of 85.38%, a sensitivity of 86.22%, and a specificity of 83.92%.

The widespread use of encrypted communication in mobile networks poses significant challenges in accurately classifying traffic. Detecting out-of-distribution (OOD) samples, which significantly deviate from known classes, adds complexity to the task. This paper proposes a feature analysis-based OOD detection scheme for traffic classification in Long-Term Evolution (LTE) systems. Our method utilizes Long Short-Term Memory (LSTM) networks for feature extraction, capturing the feature vectors of the traffic series. Principal Component Analysis (PCA) is then applied to obtain principal and residual principal components. Leveraging the residual feature vector, we construct an OOD score to quantify deviation from the ID dataset. Extensive experiments on a large-scale encrypted mobile traffic dataset demonstrate the superiority of our approach, achieving high accuracy in OOD detection compared to existing techniques. Our method contributes to enhanced security and reliable traffic classification in LTE systems, addressing challenges posed by OOD samples.

This paper addresses the sensor selection problem in time-difference-of-arrival (TDOA) localization system, where a subset of K sensors is chosen from a total of N sensors such that the trace of the Cramér-Rao lower bound (CRLB) is minimized. We present a simulated annealing (SA) based method to solve the resulting minimization problem. Simulation results demonstrate the superior performance of the proposed method compared with the previous semidefinite relaxation (SDR) based method.

To generate high-quality Synthetic Aperture Radar (SAR) images, the translation of Electro-optical (EO) to SAR images has been studied in recent years. However, these methods cannot make use of other modal information, such as Light Detection and Ranging (LiDAR), to enhance their performance, due to the absence of well-registered multi-modal remote sensing data. Therefore, in this paper, we first construct an organized high-resolution real-world benchmark dataset, which comprises 479 registered multi-modal images. We then explore how to adapt the state-of-the-art method, named the query-selected attention model (QS-Attn), to make it more suitable for integrating EO and LiDAR images into real SAR image translation. Extensive experiments have been conducted to evaluate our proposed method and compare it with state-of-the-art methods. The experimental results validate the superiority and effectiveness of our approach.

Product quantization (PQ) is a technique that transforms high-dimensional data into compact binary codes to reduce data storage and improve search efficiency. However, existing PQ methods separate the learning of modality-specific features from the learning of quantization codewords, resulting in suboptimal performance in cross-modal retrieval tasks. In this paper, we propose a joint-neighborhood product quantization (JNPQ) method to simultaneously learn modality-specific features and quantization codewords. To achieve this, we first introduce a cross-modal quantization contrastive learning module that preserves the inter-modal neighborhood of the original data and reduces the quantization error. Then, we design a self-neighbor contrastive learning module that enhances the intra-modal neighborhood within individual modalities. Extensive experiments demonstrate that JNPQ achieves state-of-the-art results in crossmodal retrieval when compared with other unsupervised crossmodal quantization methods.

This paper proposes a novel geometric solution for tracking a moving target through multistatic sensing. In contrast to existing two-step weighted least square (2SWLS) methods which use the bistatic range (BR) and bistatic range rate (BRR) measurements, the proposed method incorporates an additional direction of arrival (DOA) measurement of the target obtained from a communication receiver in an integrated sensing and communication (I SAC) system. Unlike the existing 2SWLS methods that require at least three transmitter-receiver (TX-RX) pairs to operate, the proposed algorithm can conduct location estimation with a single TX-RX pair and velocity estimation with two TX-RX pairs. Simulations reveal that the proposed method exhibits superior performance compared to existing 2SWLS methods, particularly when dealing with moderate levels of noise in DOA measurements.

Face verification has seen remarkable progress that benefits from large-scale publicly available databases. However, it remains a challenge how to generalize a pretrained face verification model to a new scenario with a limited amount of data. In many real-world applications, the training database only contains a limited number of identities with two images for each identity due to the privacy concern. In this article, we propose to transfer knowledge from a pretrained unmasked face verification model to a new model for verification between masked and unmasked faces, to meet the application requirements during the COVID-19 pandemic. To overcome the lack of intra-class diversity resulting from only a pair of masked and unmasked faces for each identity ($\text{i.e.},$two shots for each identity), a static prototype classification function is designed to learn features for masked faces by utilizing unmasked face knowledge from the pretrained model. Meanwhile, a contrastive constrained embedding function is designed to preserve unmasked face knowledge of the pretrained model during the transfer learning process. By combining these two functions, our method uses knowledge acquired from the pretrained unmasked face verification model to proceed with verification between masked and unmasked faces with a limited amount of training data. Extensive experiments demonstrate that our method can perform better than state-of-the-art methods for verification between masked and unmasked faces in the few-shot transfer learning setting.

Few-shot class incremental learning (FSCIL) aims to address catastrophic forgetting during class incremental learning in a few-shot learning setting. In this paper, we approach the FSCIL by adopting analytic learning, a technique that converts network training into linear problems. This is inspired by the fact that the recursive implementation (batch-by-batch learning) of analytic learning gives identical weights to that produced by training on the entire dataset at once. The recursive implementation and the weight-identical property highly resemble the FSCIL setting (phase-by-phase learning) and its goal of avoiding catastrophic forgetting. By bridging the FSCIL with the analytic learning, we propose a Gaussian kernel embedded analytic learning (GKEAL) for FSCIL. The key components of GKEAL include the kernel analytic module which allows the GKEAL to conduct FSCIL in a recursive manner, and the augmented feature concatenation module that balances the preference between old and new tasks especially effectively under the few-shot setting. Our experiments show that the GKEAL gives state-of-the-art performance on several benchmark datasets.

Visual Multiple Object Tracking (MOT) typically utilizes appearance and motion clues for associations. However, these features may be limited under certain challenging scenarios, such as appearance ambiguity and frequent occlusions. In this paper, we introduce a novel deep RF-affinity neural network (DRFAN) that enhances visual tracking with the aid of a passive wireless positioning device, Radio Frequency Identification (RFID). DRFAN aims to solve object tracking by introducing a new concept of a "candidate trajectory" to indicate target movement. This approach fundamentally deviates from existing fusion methods that rely on known visual tracks. Instead, DRFAN exclusively uses detection bounding boxes and RFID signals. The proposed method overcomes the limitations of visual tracking by swiftly resuming correct tracking whenever a failure occurs. This is the first time using signals from low-cost passive RFID tags to achieve image-level localization, and a discriminative neural network is designed specifically for RFID-assisted visual association. Our experimental results validate the robustness and applicability of the proposed approach.

Early detection of locomotion intention is highly relevant to the development of intelligent rehabilitation/assistive robotics. While surface electromyography(sEMG) has been a promising tool, it is often challenged by the shear variability of sEMG patterns in contrast to only a handful of sEMG training samples per discrete motion intention class for each individual user to begin with. To address this issue, we introduce a deep convolutional generative adversarial networks (DCGANs), including dynamic time warping (DTW) and fast Fourier transform mean square error (FFT MSE) for artificial signal quality assessment. On a preliminary sEMG data set of 3-class directional lower-limb movement, the proposed method yielded an average accuracy rate of$89.31 \% \pm 6.52$. While this is a feasibility study using healthy human subjects only, the result warrants extended study to further establish the generative adversarial network learning for EMG intention detection in real-world rehabilitation/assistive system applications.

Classification of traffic service types is a valuable function for wireless communication networks. Even though some progress has been made, the recognition of the type of the traffic services cannot be done in real time. In this paper, we propose a novel method for classifying traffic series in real time based on transfer learning techniques. We pre-train a deep learning model with long traffic series and fine-tune the model with short traffic series. In this way, the developed model achieves the capability of recognising traffic services in real time. In other words, the model can recognize traffic services by using short traffic series. We collect Downlink Control Information (DCI) from commercial LTE networks when using five common types of traffic services. Then we use the dataset to validate our method. Our experimental results show that, by using proposed method, LSTM accuracy rates will increase to 80% and 88.5% when the length of the traffic series is 5 seconds and 10 seconds respectively, which is higher than the baseline. The strategy is also suitable for one dimension convolution neural network (1D-CNN).

Cross-modal retrieval methods are developed to retrieve relevant data across different modalities. Usually, super-vised cross-modal retrieval methods can achieve higher accuracy than unsupervised methods because they can utilize the semantic information provided by clean labels. However, training data with noisy labels will lead to the performance degradation of supervised cross-modal retrieval methods. In this work, we present a novel framework called Neighborhood Learning for Cross-Modal Retrieval (NLCMR) that is robust against noisy labels by exploiting the information contained in the neighbor-hood. Our NLCMR contains two main components: Clustering with Neighborhood Alignment and Neighborhood Contrastive Learning. The first component focuses on reducing the impact of noisy labels and improving clustering robustness, and the second component learns from noisy data by exploring pairwise and neighborhood information. Extensive experiments are conducted on three multi-modal datasets to demonstrate the effectiveness of NLCMR.

We present a side-channel attack (SCA) resistant asynchronous-logic (async-logic) AES accelerator that integrates synchronous-logic (sync-logic) Block RAMs (BRAMs) in FPGA as the Substitution-Box. We successfully identify the timing requirements to integrate sync-logic BRAMs in our async-logic AES accelerator and validate our proposed AES accelerator on the Sakura-X FPGA board. With the integration of BRAMs, we improve the resource utilization on FPGA by$1.6\times$when compared to the state-of-the-art async-logic AES accelerator, while reducing the power overhead by$1.4\times$. We comprehensively evaluate the SCA resistance of our proposed async-logic AES accelerator with 11 attacking models in both time and frequency domains. Based on our evaluations, we show that our proposed async-logic AES accelerator is highly secure against SCA with 30 million EM traces. This is more than$6000\times$improvement when compared to the benchmark sync-logic AES accelerator and$1.5\times$improvement when compared to the state-of-the-art async-logic AES accelerator.

Clipping distortion is a common problem when the amplitude of input signals exceeds the desired region of an analog-to-digital converter. Unlimited Sensing Framework (USF) alleviates the clipping distortion by folding out-of-range signals into a within-range via modulo operations. The USF signal recovery assumes an infinitesimal residual step time in the modulo operation which is generally practically infeasible. Its recovery error is inevitable due to the remainder sampling error during the residual step transition time. Instead of the infinitesimal assumption, a residual-remainder coupled USF is proposed to eliminate the remainder sampling error by coupling the residual sampling component. It is shown that the proposed coupling framework does not only relax the oversampling rate in the original USF to approaching the Nyquist sampling rate, but also provides a more accurate signal recovery capability as the remainder sampling error is eliminated by the coupling of the residual components.

Nonnegative matrix factorization (NMF) based multiview technique has been commonly used in multiview data clustering tasks. However, previous NMF based multiview clustering approaches fail to take advantage of a small amount of supervisory information to effectively improve the clustering performance, and are easily affected by the additional post-processing method in clustering tasks. To cope with these issues, a novel framework named multiview clustering via hypergraph induced semi-supervised symmetric NMF (MVCHSS) is proposed in this paper for multiview data clustering applications. Specifically, the proposed method has the following features: 1) a new multiview based hypergraph pairwise constraints propagation (MHPCP) algorithm is developed in MVCHSS to construct a set of informative similarity matrices, revealing the high-order relationships effectively and fully utilizing the limited pairwise constraint supervisory information among samples of each view data; 2) the obtained similarity matrices with much supervisory information are not only enforced into the symmetric NMF (SNMF) model, but also incorporated into the graph regularization for each view data; 3) the optimization problem of MVCHSS is formulated for multiview data clustering tasks to acquire a more discriminative clustering indicator matrix (or called consensus assignment matrix) without additional post-processing method. Moreover, the proof of convergence and the computational complexity for MVCHSS are presented. Extensive experiments on five multiview datasets demonstrate that the proposed MVCHSS framework outperforms several state-of-the-art multiview clustering methods.

In this paper, we develop the weighted error entropy-regularized tensor learning method for multi-view subspace clustering (WETMSC), which integrates the noise disturbance removal and subspace structure discovery into one unified framework. Unlike most existing methods which focus only on the affinity matrix learning for the subspace discovery by different optimization models and simply assume that the noise is independent and identically distributed (i.i.d.), our WETMSC method adopts the weighted error entropy to characterize the underlying noise by assuming that noise is independent and piecewise identically distributed (i.p.i.d.). Meanwhile, WETMSC constructs the self-representation tensor by storing all self-representation matrices from the view dimension, preserving high-order correlation of views based on the tensor nuclear norm. To solve the proposed nonconvex optimization method, we design a half-quadratic (HQ) additive optimization technology and iteratively solve all subproblems under the alternating direction method of multipliers framework. Extensive comparison studies with state-of-the-art clustering methods on real-world datasets and synthetic noisy datasets demonstrate the ascendancy of the proposed WETMSC method.

Tracking multiple faces online in unconstrained videos is a challenging problem as faces may appear drastically different over time and identities can be inferred only based on information available from past frames. Previous tracking methods focus on face information without reference to other modality information such as a person’s overall body appearance, leading to suboptimal performance. In this paper, we propose a new online multi-face tracking method, called online multi-face tracking with multi-modality cascaded matching (OMTMCM), to improve the tracking performance by using both face and body information. The proposed OMTMCM consists of two stages, namely detection alignment and detection association. In the first stage, a detection alignment module is designed to align face detection with body detection from the same person for the subsequent detection association. In the second stage, a cascaded matching module is designed to associate face detections across frames to locate trajectory of each target face by using both face and body information. Specifically, aligned face-body detections in the current frame are matched in a cascade manner with body and face features that are selected from past frames and stored in the designed feature memory. In this way, our method can track multiple faces online with both face and body information while eliminating the possibility of face detection and body detection from the same person being separately assigned with different identities. Experimental results demonstrate our method is on par with or better than other online tracking methods for multi-face tracking.

Convolutional neural networks (CNNs) for hyperspectral image (HSI) classification have generated good progress. Meanwhile, graph convolutional networks (GCNs) have also attracted considerable attention by using unlabeled data, broadly and explicitly exploiting correlations between adjacent parcels. However, the CNN with a fixed square convolution kernel is not flexible enough to deal with irregular patterns, while the GCN using the superpixel to reduce the number of nodes will lose the pixel-level features, and the features from the two networks are always partial. In this paper, to make good use of the advantages of CNN and GCN, we propose a novel multiple feature fusion model termed attention multi-hop graph and multi-scale convolutional fusion network (AMGCFN), which includes two sub-networks of multi-scale fully CNN and multi-hop GCN to extract the multi-level information of HSI. Specifically, the multi-scale fully CNN aims to comprehensively capture pixel-level features with different kernel sizes, and a multi-head attention fusion module is used to fuse the multi-scale pixel-level features. The multi-hop GCN systematically aggregates the multi-hop contextual information by applying multi-hop graphs on different layers to transform the relationships between nodes, and a multi-head attention fusion module is adopted to combine the multi-hop features. Finally, we design a cross attention fusion module to adaptively fuse the features of two sub-networks. AMGCFN makes full use of multi-scale convolution and multi-hop graph features, which is conducive to the learning of multi-level contextual semantic features. Experimental results on three benchmark HSI datasets show that AMGCFN has better performance than a few state-of-the-art methods.

Multi-view subspace clustering aims to integrate the complementary information contained in different views to facilitate data representation. Currently, low-rank representation (LRR) serves as a benchmark method. However, we observe that these LRR-based methods would suffer from two issues: limited clustering performance and high computational cost since (1) they usually adopt the nuclear norm with biased estimation to explore the low-rank structures; (2) the singular value decomposition of large-scale matrices is inevitably involved. Moreover, LRR may not achieve low-rank properties in both intra-views and inter-views simultaneously. To address the above issues, this paper proposes the Bi-nuclear tensor Schatten- p norm minimization for multi-view subspace clustering (BTMSC). Specifically, BTMSC constructs a third-order tensor from the view dimension to explore the high-order correlation and the subspace structures of multi-view features. The Bi-Nuclear Quasi-Norm (BiN) factorization form of the Schatten- p norm is utilized to factorize the third-order tensor as the product of two small-scale third-order tensors, which not only captures the low-rank property of the third-order tensor but also improves the computational efficiency. Finally, an efficient alternating optimization algorithm is designed to solve the BTMSC model. Extensive experiments with ten datasets of texts and images illustrate the performance superiority of the proposed BTMSC method over state-of-the-art methods.

Ridge regression (RR) has been commonly used in machine learning, but is facing computational challenges in big data applications. To meet the challenges, this article develops a highly parallel new algorithm, i.e., an accelerated maximally split alternating direction method of multipliers (A-MS-ADMM), for a class of generalized RR (GRR) that allows different regularization factors for different regression coefficients. Linear convergence of the new algorithm along with its convergence ratio is established. Optimal parameters of the algorithm for the GRR with a particular set of regularization factors are derived, and a selection scheme of the algorithm parameters for the GRR with general regularization factors is also discussed. The new algorithm is then applied in the training of single-layer feedforward neural networks. Experimental results on performance validation on real-world benchmark datasets for regression and classification and comparisons with existing methods demonstrate the fast convergence, low computational complexity, and high parallelism of the new algorithm.

Radio frequency identification (RFID) is a form of wireless communication that has received much attention in recent years due to low costs of passive RFID tags and availability of commercial-off-the-shelf (COTS) RFID devices. Existing indoor localization and tracking methods based on RFID do not perform well in dynamic environments with severe multi-path interference. In this paper, we propose a robust Angle of Arrival (AoA) estimation method for mobile RFID tags in a rich multi-path environment with a large feasible area. The proposed method Tri-AoA consists of three essential modules, phase likelihood estimation, Received Signal Strength Indicator (RSSI) likelihood estimation and a deep learning algorithm. The phase likelihood estimation module exploits the concept of an antenna array to provide a basic estimation of an AoA, but with an ambiguity. The RSSI likelihood estimation module helps alleviate the ambiguity. To achieve a more robust estimation of AoA for mobile RFID tags, we construct a 2-dimensional feature image that contains AoA estimation from the phase and RSSI modules. We then develop a deep learning algorithm to analyze this image to improve the AoA tracking accuracy as well as the robustness by suppressing the multi-path interference. The experimental results show that our system outperforms existing approaches by achieving a median error of$2.36^{\mathrm{o}}$in a$3m\times 4m$area using four COTS RFID antennas. We also show that our system can realize real-time performance on a personal computer.

Unsupervised anomaly detection for time series signals is challenging, due to the imbalanced distribution of data and the lack of ground-truth labels. Current methods on this topic are mainly based on deep neural networks, which are optimized by heuristic constraints or empirical priors. However, various patterns of anomalous data, especially those lasting for varying periods, are hard to be captured by plain networks. To tackle this problem, we propose a multiple temporal context embedding method. The core of our method is to construct a unified representation of the multiple temporal contexts of data, which is achieved by learning a set of base features to reconstruct the hidden features within existing anomaly detection networks. The proposed method can be implemented as a convenient plug-in module, and be combined with various network architectures, such as autoencoders and graph neural networks. Extensive experiments on multiple datasets demonstrate that the proposed method can boost the performance of baseline networks significantly.

In this paper, a fast bearing state detection method based on multi-sensor signal fusion and compression feature extraction is proposed. The best estimation in the random weighted fusion algorithm is adaptively adjusted by the fluctuation factor to realize the high-precision fusion of variable signals and reduce the noise component in the signals. In the compressed sensing framework, a partial Hadamard matrix is selected as the measurement matrix, and the signal reconstruction is abandoned, leading to reduced average sampling rate and less data for signal acquisition, transmission, and extraction of fault features. The proposed method for diagnosis of rolling bearing fault is fast, effective, and accurate, as verified by experimental results.

In contrast to traditional adversarial learning (AL) which learns speaker-invariant representations, this paper proposes cascaded adversarial learning (CAL) which learns speaker-invariant emotion data for speaker independent emotion recognition (SIER) tasks. CAL is a dual cascaded network architecture where the output of the transformation network is fed as input to the classification network. Transformation network transforms original speech emotion to speaker-invariant emotion data by implementing an AL strategy with an encoder-decoder architecture. The classification network predicts the emotion from the speaker-invariant emotion data (output of the transformation network). We argue that the speaker-invariant emotion data realized by transformation network has less variation than the original speech emotion data and therefore are conducive for SIER as it improve generalization capability. To our knowledge this is the first time a dual cascaded network has been used for SIER and demonstrate state-of-the-art performances for SIER on Emo-DB and RAVDESS datasets.

Differential Deep Learning Analysis (DDLA) is a deep learning-based non-profiling side-channel attack leveraging neural networks to classify Physical Leakage Information with labels. To avoid the Class Imbalance Problem (CIP) of significantly different data sizes in different data groups, DDLA employs bit labels. However, applying bit labels will be less effective for exploiting leakage. In this paper, we propose to employ Correlation optimization Deep Learning Analysis (CO-DLA) to circumvent the CIP in DDLA by converting the classification in DDLA into a correlation optimization. Bus labels can then be used to exploit stronger leakage information. To validate the attack efficacy improvement, we perform experiments on ASCAD synchronized and de-synchronized masked AES-128 datasets. For the synchronized masked dataset, our proposed CO-DLA requires only 5k traces, which is 75% lesser than the 20k traces required by the reported DDLA, to reveal the key-byte. For the 2 de-synchronized masked datasets, our proposed CO-DLA requires only 10k traces to reveal the key-byte from both of them while the reported DDLA fails to reveal the key-byte.

Class-incremental learning (CIL) learns a classification model with training data of different classes arising progressively. Existing CIL either suffers from serious accuracy loss due to catastrophic forgetting, or invades data privacy by revisiting used exemplars. Inspired by learning of linear problems, we propose an analytic class-incremental learning (ACIL) with absolute memorization of past knowledge while avoiding breaching of data privacy (i.e., without storing historical data). The absolute memorization is demonstrated in the sense that the CIL using ACIL given present data would give identical results to that from its joint-learning counterpart that consumes both present and historical samples. This equality is theoretically validated. The data privacy is ensured by showing that no historical data are involved during the learning process. Empirical validations demonstrate ACIL's competitive accuracy performance with near-identical results for various incremental task settings (e.g., 5-50 phases). This also allows ACIL to outperform the state-of-the-art methods for large-phase scenarios (e.g., 25 and 50 phases).

Multi-view clustering has received widespread attention owing to its effectiveness by integrating multi-view data appropriately, but traditional algorithms have limited applicability to large-scale real-world data due to their high computational complexity and low robustness. Focusing on the aforementioned issues, we propose an efficient and robust multi-view clustering algorithm with anchor graph regularization (ERMC-AGR). In this work, a novel anchor graph regularization (ARG) is designed to improve the quality of the learned embedded anchor graph (EAG), and the obtained EAG is decomposed by nonnegative matrix factorization (NMF) under correntropy criterion to acquire clustering results directly. Different from the traditional graph regularization that needs to construct a large-scale Laplacian matrix pertaining to the all-sample graph, our lightweight AGR, constructed from the perspective of anchors, can reduce the computational complexity significantly while improving the EAG quality. Moreover, a factor matrix of NMF is constrained to be the cluster indicator matrix to omit additional k-means after optimization. Subsequently, correntropy is utilized to improve the effectiveness and robustness of ERMC-AGR owing to its promising performance to complex noises and outliers. Extensive experiments on real-world datasets and noisy datasets show that ERMC-ARG can improve the clustering efficiency and robustness while ensuring comparable or even better effectiveness.

Graph can achieve good performance to extract the low-dimensional features of hyperspectral image (HSI). However, the present graph-based methods just consider the individual information of each sample in a certain characteristic, which is very difficult to represent the intrinsic properties of HSI for the complex imaging condition. To better represent the low-dimensional features of HSI, we propose a multistructure unified discriminative embedding (MUDE) method, which considers the neighborhood, tangential, and statistical properties of each sample in HSI to achieve the complementarity of different characteristics. In MUDE, we design the intraclass and interclass neighborhood structure graphs with the local reconstruction structure of each sample; meanwhile, we also utilize the adaptive tangential affine combination structure to construct the intraclass and interclass tangential structure graphs. To further enhance the discriminating performance between different classes, we consider the influence of the statistical distribution difference for each sample to develop an interclass Gaussian weighted scatter model. Then, an embedding objective function is constructed to enhance the intraclass compactness and the interclass separability and obtain more discriminative features for HSI classification. Experiments on three real HSI datasets show that the proposed method can make full use of the structure information of each sample in different characteristics to achieve the complementarity of different features and improve the classification performance of HSI compared with the state-of-the-art methods.

Currently, illegal parking detection tasks are mainly achieved through manually checking by enforcement officers on patrol or using Closed-Circuit Television (CCTV) cameras. However, these methods either need high human labour costs or demand installation costs and procedures. Therefore, illegal parking detection solutions, which can reduce significant labour and equipment installation costs, are highly demanded. This paper proposes a novel voting based detection algorithm using deep learning networks implemented using in-vehicle cameras to achieve illegal parking detection with multiple offences’ types. Adopting in-vehicle cameras better matches real-world mobile scenarios than using traditional CCTV cameras as this helps enforcement authorities to reduce manpower and installation costs. A well-constructed new dataset with more than 10000 high-quality labelled images with seven object categories is built for illegal parking detection tasks. Additionally, one novel labelling method named “minimal illegal units” is proposed for illegal parking detection. It reduces the time and human labelling costs significantly, achieving a better correlation of a vehicle and its parking type. The experiments have been conducted in the urban areas of Singapore. Furthermore, the illumination robustness test has also been performed to illustrate that the proposed detection algorithm exhibits strong resistance to changing illumination conditions in varied operating environments. Our proposed detection algorithm can provide a benchmark for research in illegal parking detection.

Training neural networks with backpropagation (BP) requires a sequential passing of activations and gradients. This has been recognized as the lockings (i.e., the forward, backward, and update lockings) among modules (each module contains a stack of layers) inherited from the BP. In this brief, we propose a fully decoupled training scheme using delayed gradients (FDG) to break all these lockings. The FDG splits a neural network into multiple modules and trains them independently and asynchronously using different workers (e.g., GPUs). We also introduce a gradient shrinking process to reduce the stale gradient effect caused by the delayed gradients. Our theoretical proofs show that the FDG can converge to critical points under certain conditions. Experiments are conducted by training deep convolutional neural networks to perform classification tasks on several benchmark data sets. These experiments show comparable or better results of our approach compared with the state-of-the-art methods in terms of generalization and acceleration. We also show that the FDG is able to train various networks, including extremely deep ones (e.g., ResNet-1202), in a decoupled fashion.

Training neural networks with the Moore–Penrose (MP) inverse has recently gained attention in view of its noniterative training nature. However, a significant drawback of learning based on the MP inverse is that the computational memory consumption grows along with the size of a dataset. In this article, based on the partitioning of the MP inverse, we propose a blockwise recursive MP inverse formulation (BRMP) for network learning with low-memory property while preserving its training effectiveness. The BRMP is an equivalent formulation to its batchwise counterpart since neither approximation nor assumption is made in the derivation process. Our further exploration of this recursive method leads to a switching structure among three different scenarios. This structure also reveals that the well-known recursive least squares method is a special case of our proposed technique. Subsequently, we apply BRMP to the training of radial basis function networks as well as multilayer perceptrons. The experimental validation covers both regression and classification tasks.

Gradient staleness is a major side effect in decoupled learning when training convolutional neural networks asynchronously. Existing methods that ignore this effect might result in reduced generalization and even divergence. In this paper, we propose an accumulated decoupled learning (ADL), which includes a module-wise gradient accumulation in order to mitigate the gradient staleness. Unlike prior arts ignoring the gradient staleness, we quantify the staleness in such a way that its mitigation can be quantitatively visualized. As a new learning scheme, the proposed ADL is theoretically shown to converge to critical points in spite of its asynchronism. Extensive experiments on CIFAR-10 and ImageNet datasets are conducted, demonstrating that ADL gives promising generalization results while the state-of-the-art methods experience reduced generalization and divergence. In addition, our ADL is shown to have the fastest training speed among the compared methods.

The ability to simultaneously distinguish objects, materials, and their associated physical properties is one fundamental function of the sense of touch. Recent advances in the development of tactile sensors and machine learning techniques allow more accurate and complex modelling of robotic tactile sensations. However, many state-of-the-art (SotA) approaches focus solely on constructing black-box models to achieve ever higher classification accuracy and fail to adapt across sensors with unique spatial-temporal data formats. In this work, we propose an Explainable and Sensor-Adaptable Recurrent Networks (ExSARN) model for tactile texture representation. The ExSARN model consists of a two-stage recurrent networks fed by a sensor-specific header network. The first stage recurrent network emulates our human touch receptors and decouples sensor-specific tactile sensations into different frequency response bands, while the second stage codes the overall temporal signature as a variational recurrent autoen-coder. We infuse the latent representation with ternary labels to qualitatively represent texture properties (e.g. roughness and stiffness), which facilitates representation learning and provide explainability to the latent space. The ExSARN model is tested on texture datasets collected with two different tactile sensors. Our results show that the proposed model not only achieves higher accuracy, but also provides adaptability across sensors with different sampling frequencies and data formats. The addition of the crudely obtained qualitative property labels offers a practical approach to enhance the interpretability of the latent space, facilitate property inference on unseen materials, and improve the overall performance of the model.

The attack efficacy of Differential Power Analysis (DPA), a popular side channel evaluation technique for key extraction, is compromised by the false highest Difference Of Means (DOMs) value ('ghost peaks') in the DOMs matrix produced in a conventional DPA. The ghost peak is generated by the wrong key guess and always occurs in the conventional DPA when the number of side channel traces is not enough. In this paper, an improved version of the conventional DPA termed as Normalized DPA (NDPA) is proposed to circumvent the ghost peak. With the analysis on the generation of ghost peaks in the conventional DPA, we observed that by normalizing the DOMs matrix, the ghost peaks can be greatly suppressed. We model the proposed NDPA mathematically and show that it performs better than the conventional DPA. We further provide the experimental validations on a set of 200k power simulation traces on AES S- Box and 500 EM traces from ASCAD dataset. Based on the attack results of these datasets, our proposed NDPA requires (up to 68%) lesser number of traces to reveal a correct key when compared to the conventional DPA.

Filterbanks have many applications. The orthogonal wavelet filter is widely used in filterbank design because of its advantage in energy preservation. In this work, the orthogonal wavelet filter with a given length and a prescribed number of Vanishing Moment (VM) is designed to achieve minimum Root- Mean-Square (RMS) bandwidth by using a symbolic approach.

This paper proposes a kernel projection (KP) neural network that analytically determines its network parameters. The proposed network is composed of cascaded modules of 2-layer sub-networks. A technique which encodes the label information into each module has been introduced to enable a locally supervised learning. Such a supervised learning in the 2-layer module begins with a kernel projection in the first layer and determines its parameters analytically via solving a least squares problem in the second layer. We show that the analytic nature of the proposed network allows a learning process significantly faster than that of the traditional backpropagation method as it only needs to visit the dataset once. Experiments of classification tasks on various datasets are carried out, showing comparable or better results compared with several competing methods.

In this paper, we propose two normalization techniques to reduce the ghost peaks occurring in Differential Power Analysis (DPA). Ghost peaks can be defined as the DPA output generated by the wrong key guesses, having higher amplitudes than the DPA output generated by the correct key guess. We further propose variance-based Differential Power Analysis (vDPA) to attack masked crypto devices. The proposed normalization techniques and vDPA constitute four contributions. First, based on the side-channel signal modeling with the linear coefficient representing the strength of the linear component in a side-channel signal, we formulate the condition function of linear coefficients for the appearance of ghost peaks in DPA. Second, we propose pre-normalization in DPA and mathematically analyze how it can reduce ghost peaks by modulating the strength of the linear components in side-channel signals. Third, we propose post-normalization and mathematically analyze how it can reduce ghost peaks by de-correlating the strength of the linear components in side-channel signals with the condition function for the appearance of ghost peaks. Fourth, we propose vDPA to apply simultaneously with either one of the proposed normalization techniques to effectively attack masked crypto devices. Based on the experiments, we show that the proposed basic vDPA (without normalization), pre-normalized vDPA and post-normalized vDPA are all able to reveal the secret key from ASCAD data set. The pre- and post-normalized vDPAs require up to 18× and 14× fewer traces than the basic vDPA respectively. While attacking ASCAD data set, the proposed pre- and post-normalized vDPAs are both 13, 095× faster than the reported 2nd order CPA, and reveal the key-bytes successfully with only half of side-channel traces required by the reported Zero-offset DPA.

Connectionist Temporal Classification (CTC) and attention mechanism are two main approaches used in recent scene text recognition works. Compared with attention-based methods, CTC decoder has a much shorter inference time, yet a lower accuracy. To design an efficient and effective model, we propose the guided training of CTC (GTC), where CTC model learns a better alignment and feature representations from a more powerful attentional guidance. With the benefit of guided training, CTC model achieves robust and accurate prediction for both regular and irregular scene text while maintaining a fast inference speed. Moreover, to further leverage the potential of CTC decoder, a graph convolutional network (GCN) is proposed to learn the local correlations of extracted features. Extensive experiments on standard benchmarks demonstrate that our end-to-end model achieves a new state-of-the-art for regular and irregular scene text recognition and needs 6 times shorter inference time than attention-based methods.

The quantization of deep neural networks has drawn great research interest as well as industrial support for its tremendous advantages on boosting the inference speed of convolutional neural network models, especially on resource-constrained devices. Emerging research work and software framework for low-bit quantization, i.e., quantization under 8 bits, also shows strong potential. However, extreme low-bit neural network, e.g., binary neural network (BNN), suffers from significant accuracy loss. A previous work tried to improve the performance of extreme low-bit neural networks by adjusting channel numbers but achieves a poor balance between accuracy and model size. Alternatively, adjusting bitwidth for each layer gains a higher return on accuracy with an acceptable increase in model size. In this work, we employ a genetic algorithm to find the optimal bitwidth assignation for layers in VGG-Small and ResNet18. Both searched mix-bitwidth network architectures achieve higher accuracy with lower time and memory consumption, compared to their fix-bitwidth counterparts. A further investigation on the combination of BNN and the intermediate layer of higher bitwidth is also conducted.

Due to the outbreak of the novel coronavirus (or known as COVID-19), people are advised to wear masks when they stay outdoors in many countries. This could result in difficulty for some public safety surveillance systems involving face detection or tracking. Therefore, the development of face detection and tracking algorithms for people wearing face masks is particularly important. In this paper, a real-time tracking algorithm for people with or without face masks is proposed. This algorithm is trained on public face datasets with faces without masks. Although the training does not involve face images of people wearing face masks, we show that the proposed algorithm is robust as it is able to perform well in face tracking for people wearing face masks. We also discuss the possible scenarios where the algorithm could lose track of the target when experimenting in tracking masked faces. This can motivate future research in this area.

Tracking-by-detection methods track multiple objects by detecting the objects of interest in each frame and associating the detected objects with the tracks. By allowing object detection and appearance embedding to be learned in a shared network, recent tracking-by-detection methods can implement the tracking task in real time with the power of deep neural networks. However, they just focus on the detection stage and do not take advantage of the embedding features well in the association stage. In this paper, we exploit the discriminative embedding features in the association stage to improve the tracking performance. By combing the embedding features with the bounding boxes to associate the detected objects with the tracks, the number of identity switches during tracking can be reduced. Further, after associating the detected objects with the tracks, the embedding feature of each track is not only updated according to the associated object, but also learned to distinguish the similar detected objects. The experiments show that our method can achieve competitive tracking performance in real time compared to the state-of-the-art tracking methods.

Pedestrian attribute recognition aims to accurately identify attributes from pedestrian images. Visually similar pedestrian images tend to share similar attributes, and the upper and lower clothing attributes normally match in style. However, how to effectively capture these relations remains a challenging topic. In this paper, we propose a novel network Pentadent-Net (PD-Net). The network consists of a “Representation Refining Branch (RRB)” and an “Attribute Fusion Branch (AFB)”. The RRB reduces the feature distances among visually similar images, and the AFB generates a joint-representation for each attribute based on its correlation with other attributes. The polished features from the two branches are then applied to capture the inter-sample relations and attribute dependencies respectively. We also further associate features from the two branches to enrich feature representations. Experimental results on two benchmark datasets (PETA and PA-100K) by our proposed approach demonstrate its advantages over state-of-the-art methods.

A machine learning based defect detection system generally requires a new training procedure upon a new product category. In industry applications where there are variant product categories, a re-training upon category changing could be time expensive and unacceptable. In this work, a two-layer neural networks are proposed for cross-category defect detection without re-training. Different from traditional neural networks, the proposed method learns differences from image-pairs containing certain structural similarity rather than from a single image. With the assumption that different categorical objects could share certain structural similarity indicated by these learned image pairwise differences, a pairwise Siamese neural network is used in the proposed neural networks for defect detection. The cross-category capability of the proposed method is evidenced via experiments based on real-world factory datasets.

Pedestrian Attribute Recognition aims to recognise person attributes including age, gender, clothing and accessories in a given image. It is challenging due to the high variance in sample quality and attribute content. Many network structures have been proposed to better capture the fine-grained details in an image. However, how to learn from each training sample based on their importance to the model still remains to be addressed. In this paper, we propose Reinforced Sample Re-weighting (RSR), a novel approach to re-weight samples in a batch during back-propagation through reinforcement learning. RSR agents are proposed to assign sample weights based on both the sample itself and the recognition model status. The agent learns in an on-policy manner, where it learns together with the attribute recognition model and no additional training is required. The proposed approach achieves state-of-the-art performance against other existing methods on three large scale pedestrian attribute datasets PETA, PA-100K and RAP, which demonstrates the effectiveness of our method.

This paper presents a comprehensive study on radar detection using deep learning with application to automotive vehicles. Automotive radars face complex target scenarios consisting of both small point targets and large extended targets. However, the current works on automotive radar detection mainly focus on point target detection. Moreover, those works use the complex range Doppler data for detection. In this paper, a deep learning based method for extended target detection is presented that takes advantage of augmented data for neural network training and prediction. Extensive simulations have been conducted to evaluate the proposed detection method and the results show performance improvement over a recent related method.

The sense of touch is an essential sensing modality for a robot to interact with the environment as it provides rich and multimodal sensory information upon contact. It enriches the perceptual understanding of the environment and closes the loop for action generation. One fundamental area of perception that touch dominates over other sensing modalities, is the understanding of the materials that it interacts with, for example, glass versus plastic. However, unlike the senses of vision and audition which have standardized data format, the format for tactile data is vastly dictated by the sensor manufacturer, which makes it difficult for large-scale learning on data collected from heterogeneous sensors, limiting the usefulness of publicly available tactile datasets. This paper investigates the joint learnability of data collected from two tactile sensors performing a touch sequence on some common materials. We propose a supervised recurrent autoencoder framework to perform joint material classification task to improve the training effectiveness. The framework is implemented and tested on the two sets of tactile data collected in sliding motion on 20 material textures using the iCub RoboSkin tactile sensors and the SynTouch BioTac sensor respectively. Our results show that the learning efficiency and accuracy improve for both datasets through the joint learning as compared to independent dataset training. This suggests the usefulness for large-scale open tactile datasets sharing with different sensors.

Visual attention mechanism extracts features by establishing a mask branch characterizing the distribution regions we are interested in. With the thought that eliminating useless features helps in the process of extracting significant features, in this paper we design a novel visual attention mechanism structure called the mixed attention module (MAM), which is capable of distilling significant patterns as well as suppressing unimportant features. The generic MAM constructed by the min-pooling and the tanh function is embedded into a mixed attention network for image classification. Moreover, a new image enhancement method named the random disruption (RandomDisrupt) is developed by slicing an image and then scrambling it. RandomDisrupt destroys the positional information between the features and the structural information of the object to accomplish data augmentation. The combination of the MAM and RandomDisrupt has improved the image classification performance significantly. Benchmark experiments are conducted to show the superiority of the proposed algorithm over a state-of-the-art method. Source code would be available at https://github.com/weiyu-zeng/MAM.

Nonnegative matrix factorization (NMF) has attracted increasing attention in data mining and machine learning . However, existing NMF methods have some limitations. For example, some NMF methods seriously suffer from noisy data contaminated by outliers, or fail to preserve the geometric information of the data and guarantee the sparse parts-based representation. To overcome these issues, in this paper, a robust and sparse NMF method, called correntropy based dual graph regularized nonnegative matrix factorization with local coordinate constraint (LCDNMF) is proposed. Specifically, LCDNMF incorporates the geometrical information of both the data manifold and the feature manifold, and the local coordinate constraint into the correntropy based objective function. The half-quadratic optimization technique is utilized to solve the nonconvex optimization problem of LCDNMF, and the multiplicative update rules are obtained. Furthermore, some properties of LCDNMF including the convergence, relation with gradient descent method , robustness, and computational complexity are analyzed. Experiments of clustering demonstrate the effectiveness and robustness of the proposed LCDNMF method in comparison to several state-of-the-art methods on six real world image datasets.

One of the salient features of the extreme learning machine (ELM) is its fast learning speed. However, in a big data environment, the ELM still suffers from an overly heavy computational load due to the high dimensionality and the large amount of data. Using the alternating direction method of multipliers (ADMM), a convex model fitting problem can be split into a set of concurrently executable subproblems, each with just a subset of model coefficients. By maximally splitting across the coefficients and incorporating a novel relaxation technique, a maximally split and relaxed ADMM (MS-RADMM), along with a scalarwise implementation, is developed for the regularized ELM (RELM). The convergence conditions and the convergence rate of the MS-RADMM are established, which exhibits linear convergence with a smaller convergence ratio than the unrelaxed maximally split ADMM. The optimal parameter values of the MS-RADMM are obtained and a fast parameter selection scheme is provided. Experiments on ten benchmark classification data sets are conducted, the results of which demonstrate the fast convergence and parallelism of the MS-RADMM. Complexity comparisons with the matrix-inversion-based method in terms of the numbers of multiplication and addition operations, the computation time and the number of memory cells are provided for performance evaluation of the MS-RADMM.

Design of infinite impulse response (IIR) digital filters to approximate some desired magnitude-frequency response is a classical research topic in signal processing. When a pole radius constraint is imposed, however, the problem becomes challenging and few solution methods are available. This paper converts the magnitude-response approximation problem into another problem that approximates the desired magnitude response and an accompanied phase response simultaneously. By iteratively updating the accompanied phase response, a solution to the original magnitude-response approximation problem can be obtained. A striking feature of the proposed method is that the pole radius constraint can be easily incorporated in the problem. Simulations and comparisons demonstrate the effectiveness of the method.

Classification of non-tumorous facial pigmentation disorders is an important but overlooked problem. Recently, a voting-based probabilistic linear discriminant analysis (V-PLDA) method was developed to address this problem by extracting hand-craft features from a given image set of rather small size, with limited classification accuracy. In this paper, we propose an improved Synthetic Minority Over-sampling Technique (improved SMOTE) with several parameters tuned to fully utilize the available images. Moreover, transfer learning is applied to reduce the data size requirement of the deep learning model. By combining the improved SMOTE and transfer learning, a classification accuracy gain (10%) is attained compared to the state-of-the-art V-PLDA method.

Recently, a learning method based on the kernel and the range space projections has been introduced. This method has been applied to learn the multilayer network analytically with interpretable relationships among the weight matrices. However, the learning method carries a high-memory demand during training. In this study, a low-memory formulation is proposed to address this issue of high-memory demand. The developed method is inspired by a recursive implementation of the Moore-Penrose inverse and is shown to be mathematically equivalent to the original batch learning. Next, we further improved our proposed low-memory formulation to annul the potential divergence caused by rounding errors. The regression and classification behaviors of the proposed learning method are demonstrated using both synthetic and benchmark datasets. Our experiments confirm that the proposed formulation consumes significantly lower memory.

Non-tumorous facial pigmentation, though not fatal, adversely affects one's quality of life and may indicate concurrence of systemic diseases. Automatic diagnosis method such as voting-based probabilistic discriminant analysis (V-PLDA) has been explored, but the accuracy of classification is not satisfactory due to the limited number of data for training. This paper proposes to use the pre-trained deep learning network of Inception-ResNet-v2 so that information from similar datasets can be utilized. Furthermore, data augmentation using synthetic minority over-sampling technique (SMOTE) is also applied to make full use of available training data. A clinical dataset of five most common types of non-tumorous facial pigmentation disorders in Asia, namely freckles, lentigines, melasma, Hori's nevus, and nevus of Ota, is used for training and testing. The classification accuracy has shown significant improvement (> 7%) compared to the state-of-the-art method.

Nonnegative matrix factorization (NMF) has been a state-of-the-art data representation method, since it contains the psychological and physiological evidence for parts-based representation in the human brain. However, many existing NMF methods fail to ensure the decomposed results to be sparse, or ignore some useful geometrical structure information in the data. In this paper, a sparse NMF method, called dual graph regularized nonnegative matrix factorization with l1-norm sparsity constraint (l1-DNMF) is proposed to solve the two problems together. In addition, to satisfy the locality condition and sparsity constraint simultaneously, we also propose the dual graph regularized nonnegative matrix factorization with local coordinate constraint (LDNMF). By using the multiplicative update algorithm to solve the optimization problems of l1-DNMF and LDNMF, we derive two efficient alternating iterative methods. Experimental results on four image datasets demonstrate the promising performance of the new methods compared with several related methods for clustering applications.

Due to its small size, slow flying speed, and low flying altitude, classification of mini-sized unmanned aerial vehicles (UAVs) using a frequency-modulated continuous wave (FMCW) surveillance radar is a challenging task. This is because the FMCW radar echo signals are acquired at a short dwell time and thus contain limited information about targets. In this paper, we first analyze FMCW radar returns from various types of UAVs and non-UAV objects in terms of the micro-Doppler signature (m-DS) pattern. Based on the analysis results, we propose an effective and efficient UAV classification system using FMCW radar echo signals. The proposed system consists of five main parts namely, (i) burst selection, (ii) rule-based scan pruning, (iii) the empirical mode decomposition based m-DS analysis and features extraction, (iv) error counting minimization based class label estimation, and (v) scan-to-scan filtering. Our experimental results on physically measured FMCW radar echo signals from several types of UAVs and non-UAV objects show that the proposed system consistently outperforms a commercial-off-the-shelf UAV classification system in terms of the classification accuracy .

In this study, the authors propose an efficient extension of the standard empirical mode decomposition (EMD) for complex‐valued univariate signal decomposition. The key idea of the extension is to convert a complex‐valued univariate signal into a longer real‐valued signal by augmenting the real part with the flipped imaginary part, and then to decompose it into intrinsic mode functions (IMFs) using the EMD once only. The bivariate IMFs are then retrieved from the obtained IMFs. Their empirical results on synthetic data show that the proposed method significantly outperforms the traditional bivariate EMD (BEMD) method in terms of computational efficiency while producing a comparable extraction error. Moreover, the proposed method shows better micro‐Doppler signature analysis performance on physically measured continuous‐wave radar data than that of the BEMD.

This letter presents a class of infinite impulse response (IIR) filters synthesized using a system of IIR subfilters based on the frequency-response masking (FRM) technique. In the proposed class of filters, the bandedge shaping filters are a parallel connection of two allpass filters, and the masking filters are approximately linear phase IIR filters composed of a parallel connection of a delay line and an allpass filter. A design method which jointly optimizes the IIR subfilters is introduced. The proposed class of IIR filters has lower coefficient sensitivity and roundoff noise compared with conventional IIR filters.

The extreme learning machine (ELM) has attracted much attention over the past decade due to its fast learning speed and convincing generalization performance. However, there still remains a practical issue to be approached when applying the ELM: the randomly generated hidden node parameters without tuning can lead to the hidden node outputs being nonuniformly distributed, thus giving rise to poor generalization performance. To address this deficiency, a novel activation function with an affine transformation (AT) on its input is introduced into the ELM, which leads to an improved ELM algorithm that is referred to as an AT-ELM in this paper. The scaling and translation parameters of the AT activation function are computed based on the maximum entropy principle in such a way that the hidden layer outputs approximately obey a uniform distribution. Application of the AT-ELM algorithm in nonlinear function regression shows its robustness to the range scaling of the network inputs. Experiments on nonlinear function regression, real-world data set classification, and benchmark image recognition demonstrate better performance for the AT-ELM compared with the original ELM, the regularized ELM, and the kernel ELM. Recognition results on benchmark image data sets also reveal that the AT-ELM outperforms several other state-of-the-art algorithms in general.

In this paper, we first investigate into six popular entropies extracted from a set of intrinsic mode functions (IMFs) as a feature pattern for radar-based mini-size unmanned aerial vehicles (mini-UAV) classification. The six entropies include Shannon entropy, spectral entropy, log energy entropy, approximate entropy, fuzzy entropy and permutation entropy. Via an empirical comparison among the six entropies on real measurement radar data, the first three are selected as the representative due to their high efficiency and accuracy. To enhance the classification accuracy, the three selected entropies are then extracted from eight different sets of IMFs obtained by signal downsampling, and then fused at feature level. The nonlinear support vector machine classifier is adopted to predict the class label of unseen test radar signals. Our empirical results on a set of real-world continuous wave radar data show that the proposed method outperforms the state-of-the-art method in terms of the mini-UAV classification accuracy.

Heart disease is one of the leading causes of mortality worldwide. Cardiac rehabilitation plays a crucial role in enhancing quality of life and helping cardiac patients in returning to functional baseline. This paper studies cardiac patients during rehabilitation exercises by monitoring their ECG signals. A novel minimum error entropy based R peak detection scheme is proposed. ECG signals from different types of rehabilitation exercises are processed and analyzed. Unlike traditional methods which rely on intermittent measurements, the described ECG based rehabilitation monitoring provides continuous and detailed information about patients' cardiac activities. The findings from this study will be helpful in the personalized cardiac rehabilitation.

In this paper, we address the important problem of direction of arrival (DOA) estimation using a single directional antenna under the multipath environment. We address this practically important problem by following a three step approach; i) Restoration, ii) Separation and iii) Estimation. By assuming constant modulus signal similar to earlier works, we employ the constant modulus algorithm (CMA) to restore the constant modularity which gets destroyed due to multipath, in the restoration step. In the subsequent separation step, using the restored signal as the reference, we separate the power profile corresponding to each of the direct and reflected paths. Further, to estimate DOA from these separated power profiles, a compressed sensing based algorithm is also proposed. Both the simulation and experimental results show significantly better multipath mitigation with the proposed approach than those by the existing approaches.

One of the most attractive features of the extreme learning machine (ELM) is its fast speed of learning. In a big data environment, however, ELM may still suffer an overly-heavy computational issue. This paper presents a novel relaxed alternating direction method of multipliers (ADMM) for convex model fitting problems with a focus on a highly parallel implementation for least-squares problems arising from neural network training by ELM. Convergence results and computational complexity of the relaxed ADMM for least-squares problems are given, and comparisons with existing methods are also provided.

Deep convolutional neural networks (CNNs) have become one of the state-of-the-art methods for image classification in various domains. For biomedical image classification where the number of training images is generally limited, transfer learning using CNNs is often applied. Such technique extracts generic image features from nature image datasets and these features can be directly adopted for feature extraction in smaller datasets. In this paper, we propose a novel deep neural network architecture based on transfer learning for microscopic image classification. In our proposed network, we concatenate the features extracted from three pretrained deep CNNs. The concatenated features are then used to train two fully-connected layers to perform classification. In the experiments on both the 2D-Hela and the PAP-smear datasets, our proposed network architecture produces significant performance gains comparing to the neural network structure that uses only features extracted from single CNN and several traditional classification methods.

Nonnegative matrix factorization (NMF) is a significant matrix decomposition technique for learning parts-based, linear representation of nonnegative data, which has been widely used in a broad range of practical applications such as document clustering, image clustering, face recognition and blind spectral unmixing. Traditional NMF methods, which mainly minimize the square of the Euclidean distance or the Kullback-Leibler (KL) divergence, seriously suffer the outliers and non-Gaussian noises. In this paper, we propose a robust sparse nonnegative matrix factorization algorithm, called l1-norm nonnegative matrix factorization based on maximum correntropy criterion (11-CNMF). Specifically, l1-CNMF is derived from the traditional NMF algorithm by incorporating the l1 sparsity constraint and maximum correntropy criterion. Numerical experiments on the Yale database and the ORL database with and without apparent outliers show the effectiveness of the proposed algorithm for image clustering compared with other existing related methods.

The optimal design of two-channel FIR quadrature mirror filter (QMF) banks based on the Frequency-Response-Masking (FRM) technique is studied in this paper. An iterative algorithm is proposed to design a prototype filter. In each iteration of the algorithm, the prototype filter is optimized in the constrained weighted minimax sense with the weighting function being updated for reducing the reconstruction error of the filter banks. Further, we give an alternating variable approach to solving these constrained weighted minimax design problems of FRM filters. Specifically, it converts the nonlinear best approximation problem into a series of linear best approximation subproblems and solves each one using an efficient Goldfarb-Idnani based algorithm. The proposed filter banks have lower overall arithmetic complexity compared with the conventional ones. An example is presented and the simulation results show the proposed method performs better than an existing method.

In this letter, we propose an empirical-mode decomposition (EMD)-based method for automatic multicategory mini-unmanned aerial vehicle (UAV) classification. The radar echo signal is first decomposed into a set of oscillating waveforms by EMD. Then, eight statistical and geometrical features are extracted from the oscillating waveforms to capture the phenomenon of blade flashes. After feature normalization and fusion, a nonlinear support vector machine is trained for target class-label prediction. Our empirical results on real measurement of radar signals show encouraging mini-UAV classification accuracy performance.

Wavelet filters with rational coefficients can be implemented efficiently in digital hardware using simple register shifts and adders. In this letter, a technique to construct biorthogonal filters with rational coefficients is presented. The filters have linear phase and also have very similar characteristics to orthogonal filters, while the latter cannot have linear phase. The frame bound ratio of the filters is very close to unity, i.e., almost tight. The filters are constructed via the lifting scheme with four lifting steps. The perfect reconstruction and the no-dc-leakage properties are preserved with these rational coefficients filters.

Gabor magnitude is known to be among the most discriminative representations for face images due to its space- frequency co-localization property. However, such property causes adverse effects even when the images are acquired under moderate head pose variations. To address this pose sensitivity issue and other moderate imaging variations, we propose an analytic Gabor feedforward network which can absorb such moderate changes. Essentially, the network works directly on the raw face images and produces directionally projected Gabor magnitude features at the hidden layer. Subsequently, several sets of magnitude features obtained from various orientations and scales are fused at the output layer for final classification decision. The network model is analytically trained using a single sample per identity. The obtained solution is globally optimal with respect to the classification total error rate. Our empirical experiments conducted on five face data sets (six subsets) from the public domain show encouraging results in terms of identification accuracy and computational efficiency.

Filterbanks are widely used in many applications. In the literature, most filterbanks with irrational value coefficients were designed, while in practice, filters and filterbanks are implemented using rational coefficients. This paper introduces a powerful mathematical tool called Gröbner basis (GB)into the design of orthogonal filterbanks. Specifically, we use GB to find good rational lattice parameters for Daubechies length 6 (Db6) filterbank. We show that there does not exist any Db6 filterbank with rational coefficients and also meeting the first two vanishing moments (VM's). Moreover, we obtain several good Db6 filterbanks which meet the first VM exactly and the second VM approximately.

Document image binarization is a crucial step towards optical character recognition and analysis. One common way to achieve image binarization is thresholding. Thresholding methods can be divided into global and local ones in terms of the regional information used in obtaining the threshold values. Both methods have their respective drawbacks. Global methods can not adapt to background variations while local methods have the problem of local widow size determination. Hybrid methods that combines both local and global thresholds and can alleviating these drawbacks. In this paper, the hybrid threshold method is utilized and trade-off between the local and global contents is determined using variational optimization. The proposed algorithm is tested on (H-)DIBCO benchmarks and has shown superior performance to one state-of-the-art document image binarization method.

This paper proposes a method based on the bag-of-words (BoW) and the softmax regression for microscopic image classification. Essentially, the locality-constrained linear coding (LLC) is adopted for local feature encoding. Compared with the traditionally adopted vector quantization (VQ) in the BoW framework, the LLC encodes local structures of microscopic images with lower quantization errors and generates a sparse image representation. This enables the use of linear classifiers with low computational complexity. A softmax regression classifier is then adopted to address the multi-categorical classification task where the confidence of categorical prediction is quantified by posterior probabilities. Compared with other linear classifiers (such as the linear SVM) which only assign labels to images, such probabilistic outputs provide extra quantitative information to analyze misclassified images. Our experiments on the 2D-Hela and the PAP smear data sets show significant performance improvement of the proposed method comparing with competing methods using different features and classifiers under the BoW framework.

This paper proposes a modified spatially-constrained similarity measure (mSCSM) method for endosomal structure detection and localization under the bag-of-words (BoW) framework. To our best knowledge, the proposed mSCSM is the first method for fully automatic detection and localization of complex subcellular compartments like endosomes. Essentially, a new similarity score and a novel two-stage output control scheme are proposed for localization by extracting discriminative information within a group of query images. Compared with the original SCSM which is formulated for instance localization, the proposed mSCSM can address category based localization problems. The preliminary experimental results show the proposed mSCSM can correctly detect and localize 79.17% of the existing endosomal structures in the microscopic images of human myeloid endothelial cells.

IIR frequency-response masking (FRM) filter where the prototype filter is an IIR filter and the masking filters are FIR filters is effective in reducing the group delay of FRM filters. In this paper, we present a constrained optimization method for the design of near linear phase IIR FRM filters, where the phase error and magnitude error are independently controlled. A design example is presented to demonstrate the performance of the method.

Constrained adaptive filtering algorithms have been extensively studied in many applications. Most existing constrained adaptive filtering algorithms are developed under the mean square error (MSE) criterion, which is an ideal optimality criterion under Gaussian noises. This assumption however fails to model the behavior of non-Gaussian noises found in practice. Motivated by the robustness and simplicity of maximum correntropy criterion (MCC) for non-Gaussian impulsive noises, this paper proposes a new adaptive filtering algorithm called constrained maximum correntropy criterion (CMCC). Specifically, CMCC incorporates a linear constraint into a MCC filter to solve a constrained optimization problem explicitly. The proposed adaptive filtering algorithm is easy to implement, has low computational complexity, and can significantly outperform those MSE based constrained adaptive algorithms in heavy-tailed impulsive noises. Additionally, the mean square convergence behaviors are studied under energy conservation relation, and a sufficient condition to ensure the mean square convergence and the steady-state mean square deviation (MSD) of the CMCC algorithm are obtained. Simulation results confirm the theoretical predictions under both Gaussian and non-Gaussian noises, and demonstrate the excellent performance of the novel algorithm by comparing it with other conventional methods.

Based on the knowledge that input data distribution is important for learning, a data density-dependent quantization scheme (DQS) is proposed for sparse input data representation. The usefulness of the representation scheme is demonstrated by using it as a data preprocessing unit attached to the well-known least squares support vector machine (LS-SVM) for application on big data sets. Essentially, the proposed DQS adopts a single shrinkage threshold to obtain a simple quantization scheme, which adapts its outputs to input data density. With this quantization scheme, a large data set is quantized to a small subset where considerable sample size reduction is generally obtained. In particular, the sample size reduction can save significant computational cost when using the quantized subset for feature approximation via the Nyström method. Based on the quantized subset, the approximated features are incorporated into LS-SVM to develop a data density-dependent quantized LS-SVM (DQLS-SVM), where an analytic solution is obtained in the primal solution space. The developed DQLS-SVM is evaluated on synthetic and benchmark data with particular emphasis on large data sets. Extensive experimental results show that the learning machine incorporating DQS attains not only high computational efficiency but also good generalization performance.

People today rely more and more on global positioning system (GPS) for navigation when driving in unfamiliar environments. While GPS navigation is indispensable in an intelligent vehicle and provides convenience for road direction, concerns are also raised if the use of GPS may distract drivers to increase unnecessary stress. In this paper, we explore the effects of using GPS navigation on driver stress utilizing electrocardiogram (ECG) signals. In particular, the effects of higher or lower density of GPS instructions are studied. To analyze the driver stress, eight heart rate variability (HRV) features, which were commonly utilized in human stress related studies, were computed from ECG signals. Statistical significance tests were then performed to each HRV feature, so that those effective features for detecting driver stress may be localized. Our studies, based on road driving experiments with ten healthy subjects, showed that MeanRR, SDNN and HRVTri are the top three effective features to detect driver stress, while frequency domain features in general are not sensitive to driver stress. Based on the effective features, our analysis further showed that basically, driving with higher density of GPS instructions has no significant driver stress difference from driving with lower density of GPS instructions.

This paper proposes a new method for melasma pigmentary area segmentation utilizing re action-diffusion based level set model (RDLSM) together with local entropy thresholding. In the adopted level set model, a diffusion term is used to regularize the level set function while a reaction term with anticipated sign property is used to force the zero level set towards desired locations. Then local entropy thresholding is applied to address the over-segmentation issue of RDLSM and to extract desired boundaries with higher overall local entropy. As a result, the melasma pigmentary areas and the normal skin areas can be better identified. Experimental results show that the proposed method performs well for melasma image segmentation, especially for cases with severe non-uniform illumination distribution.

Independent component analysis (ICA) by an information measure has seen wide applications in engineering. Different from traditional probability density function based information measures, a probability survival distribution based Cauchy-Schwartz information measure for multiple variables is proposed in this paper. Empirical estimation of survival distribution is parameter-free which is inherited by the estimation of the new information measure. This measure is proved to be a valid statistical independence measure and is adopted as an objective function to develop an ICA algorithm which is validated by an experiment. This work shows promising potential regarding the use of survival distribution based information measure for ICA.

Monitoring the presence of occupants in a room in a timely manner is a fundamental step for effective building management. Environmental sensor networks have the advantages of high cost-efficiency and non-intrusiveness on privacy and are very suitable for room occupancy detection. Nonlinear discriminative models, e.g., support vector machine and neural networks, have shown good detection performance due to their ability to model complex relationship. However, they tend to produce unstable detection with frequent fluctuations over time, because they regard training data as independent and ignore the prior knowledge of the room occupancy, i.e., not changing very frequently. To improve the stability of the detection, we propose a two-stage structured learning approach with Extreme Learning Machine (ELM) as the local classifier. In the first stage, ELM is used as a fast nonlinear classifier to obtain preliminary detection results. In the second stage, we form data sequences consisting of the current and previous data points. The preliminary detection results by ELM of the data sequences are then used as input to a linear support vector machine for structured output to generate the final detection results. We test the proposed two-stage structured learning approach on a real-world dataset and show that the proposed approach outperforms the related machine learning methods.

A new class of filter banks, that have structural perfect reconstruction, and its design is presented here. The filters are defined by four halfband filters and are constructed using the lifting scheme. The filter banks here is a generalization of the triplet halfband filter banks proposed by Ansari et. al. (1999). Filters that have linear phase response (symmetric coefficients) and are also virtually energy preserving can be constructed using the class of structure proposed.

Real-time driver distraction detection is the core to many distraction countermeasures and fundamental for constructing a driver-centered driver assistance system. While data-driven methods demonstrate promising detection performance, a particular challenge is how to reduce the considerable cost for collecting labeled data. This paper explored semi-supervised methods for driver distraction detection in real driving conditions to alleviate the cost of labeling training data. Laplacian support vector machine and semi-supervised extreme learning machine were evaluated using eye and head movements to classify two driver states: attentive and cognitively distracted. With the additional unlabeled data, the semi-supervised learning methods improved the detection performance (G-mean) by 0.0245, on average, over all subjects, as compared with the traditional supervised methods. As unlabeled training data can be collected from drivers' naturalistic driving records with little extra resource, semi-supervised methods, which utilize both labeled and unlabeled data, can enhance the efficiency of model development in terms of time and cost.

Compressed sensing (CS) has been widely researched in the past decade due to its important contributions in sparse signal processing. In this paper, we study the problem of multiple hypotheses classification with sparse signals in compressed sensing. The performance of classifying sparse signals reconstructed with the underdetermined linear measurements under Gaussian random noise is considered. With the prior knowledge of the support set of a sparse signal, the theoretical classification bound with the recovered signal based on the oracle estimator and the restricted isometry property (RIP) of the sampling matrix is developed. The effectiveness of the proposed theoretical bound is demonstrated by the simulations results obtained by four representative reconstruction algorithms in CS.

Sixteen conventional heart beat variability (HRV) parameters and eight vital signs have shown promise in the prediction of cardiac arrest within 72 hours. Besides these 24 parameters, we proposed adding two new features for cardiac arrest prediction, which are approximate entropy (ApEn) and sample entropy (SpEn). ApEn and SpEn are nonlinear HRV parameters capable of characterizing heart conditions. These two entropies were derived from electrocardiography recordings and combined with the existing 24 features to form feature combinations. The experiments were conducted by using linear kernel Support Vector Machine classification technique to investigate the effects of using ApEn, SpEn together with 24 parameters on cardiac arrest prediction. The dimensionality reduction approach, Principal Component Analysis, was applied to suppress the dimensionality. Results reveal that the prediction performance of adding ApEn and SpEn to the 24 parameters is improved significantly compared to using the 24 parameters only. Dimension reduction has additional positive effects on improving the prediction results.

Biomolecular interactions are central to biological processes and typically take place at nanometer scale distances. They often involve molecular motion which is known to affect the accuracy of the parameter estimates. Therefore, in this paper, we consider a case of two closely spaced molecules with planar trajectory and present a general expression of the Fisher information matrix in terms of their trajectory from which the benchmark for the accuracy of the parameter estimates is obtained. Through simulations, we show its application in the case of two moving objects and another case where only one of the two objects is moving. It is shown that the deterioration of the limit of the accuracy is not only dependent on the proximity of their starting position but also on their speed and direction of movement. The effect of differing photon emission intensities on the limit of the accuracy of parameter estimation is also investigated.

In this paper, we present an overview of recent advances in multidimensional (MD) systems and signal processing. We focus on topics closely related to the four papers selected into the special session of “recent advances in multidimensional systems and signal processing” at ISCAS 2015. The paper starts with an overview of the theory of MD IIR digital filters and its applications ranging from image/videos of light-fields to microwave and mm-wave antenna array processing. State-space formulation for the realization of MD IIR notch filters is also discussed as applicable to image processing scenarios. Thereafter, new developments in visual tomography-based imaging systems that exploit MD signal processing towards safety and health applications are discussed. The paper also reviews new theoretical developments in modeling of physical systems. Recent advances in MD Kirchhoff circuit realizations for some physical systems including finite speed heat diffusion are reviewed.

In this paper, a voting based weighted online sequential extreme learning machine (VWOS-ELM) is proposed for class imbalance learning (CIL). VWOS-ELM is the first sequential classifier that can tackle the class imbalance problem in multi-class data streams. Utilizing WOS-ELM and the recently proposed voting based online sequential extreme learning machine (VOS-ELM) method, VWOS-ELM adapts better to newly received data than the original WOS-ELM method. Experimental results show that VWOS-ELM outperforms both the WOS-ELM and the recent meta-cognitive extreme learning machine methods. It also achieves similar performance to that of ensemble of subset OS-ELM (ESOS-ELM) but using fewer independent classifiers.

Graph signal processing is an emerging area that has a wide variety of applications, e.g. energy networks, transportation networks and neuronal networks. Narang and Ortega (2012) proposed the critically sampled two-channel filter bank for signals on undirected graphs using spectral graph theory. The design of graph QMF (Quadrature-Mirror-Filters) by Narang and Ortega (2012) is based on the Chebyshev polynomial approximation of the ideal Meyer function. The reconstruction error using this method can be quite large especially for low degree filters and the high-pass filters suffer from DC leakage. A simple method that is analytically based is presented here for the design of the graph QMF without DC leakage and with low reconstruction error. The filters have flat spectral response at the DC and aliasing frequencies.

We propose a novel real-time system design for multiband signal blind reconstruction using multi-coset sampling theory. Multi-channel signals are acquired under sub-Nyquist sampling frequency to perfectly reconstruct the original signal spectrum. A novel system design with Field-Programmable Gate Array (FPGA) implementation is presented in this paper. There are two main contributions in this paper. Firstly, the FPGA system uses 32-bit single precision floating point dataflow rather than conventional 16-bit fixed point to recover signals with much lower Signal-Noise Ratio (SNR). Secondly, we introduce a novel Jacobi CORDIC eigenvalue decomposition (EVD) core using parallel pivot-seeking circuit and parallel 3-CORDIC design to improve speed significantly. Hermitian matrices of dimensions from 2 to 10 are tested to compare conventional 2-CORDIC EVD and proposed EVD. The proposed EVD effectively reduces on average 36% of processing time for mesh connection system and over 50% for parallel system.

The processing of signal on graphs is becoming an important emerging area that has great potential in a wide variety of applications, e.g. social network. The work by Narang and Ortega (2012) laid the framework for critically sampled orthogonal two-channel filter bank for signal on undirected bipartite graphs. The design method presented by Narang and Ortega (2012) does not allow the tailoring of the filters spectral response and the control of the reconstruction error of the filter bank. This work present a method to design spectral filters that is based on Bernstein polynomial approximation and constrained optimization. The method allows the trade-off between transition band sharpness, ripple magnitude and reconstruction error.

Extreme learning machine (ELM) has recently attracted many researchers' interest due to its very fast learning speed, good generalization ability, and ease of implementation. It provides a unified solution that can be used directly to solve regression, binary, and multiclass classification problems. In this paper, we propose a stacked ELMs (S-ELMs) that is specially designed for solving large and complex data problems. The S-ELMs divides a single large ELM network into multiple stacked small ELMs which are serially connected. The S-ELMs can approximate a very large ELM network with small memory requirement. To further improve the testing accuracy on big data problems, the ELM autoencoder can be implemented during each iteration of the S-ELMs algorithm. The simulation results show that the S-ELMs even with random hidden nodes can achieve similar testing accuracy to support vector machine (SVM) while having low memory requirements. With the help of ELM autoencoder, the S-ELMs can achieve much better testing accuracy than SVM and slightly better accuracy than deep belief network (DBN) with much faster training speed.

A weighted least squares scheme based on an empirical survival error potential function is proposed in this paper. The empirical survival error potential function provides an error compensation scheme for noise distributions far from being Gaussian. This error compensation procedure is efficiently implemented via a weighted least squares formulation where an analytical solution form is obtained. The performance of the developed scheme is extensively tested on 16 benchmark data sets where the results show promising potential of the proposed empirical survival error distribution compensation scheme for binary pattern classification.

This paper proposes a weighted least squares (WLS) optimization procedure for the design of infinite impulse response (IIR) filters and IIR two-channel quadrature mirror filter banks (FB). The original WLS design problem is converted to a sequence of WLS sub-problems. In each of the WLS sub-problems, only one second-order factor of the denominator and the numerator is updated and iteratively solved by a constrained WLS algorithm. Using the same IIR filter from a recent work, the proposed method shows a better performance in terms of all the measured errors. For the FB, the lowpass filter is designed using a ladder structure through an allpass filter and desirable performance is achieved using a low order allpass filter.

For a time-limited sequence, the Root-Mean-Square (RMS) bandwidth is the normalized second moment of the spectrum. The RMS bandwidth is a useful and analytically tractable measure of the frequency localization of a discrete-time filter. In this work the design of orthogonal wavelet filter banks with a prescribed number of Vanishing Moment (VM) and having a minimum RMS bandwidth is considered. It is shown that the design problem can be cast as a convex optimization problem for which efficient algorithms and software for its solution exist.

Fast and accurate risk stratification is essential in the emergency department (ED) as it allows clinicians to identify chest pain patients who are at high risk of cardiac complications and require intensive monitoring and early intervention. In this paper, we present a novel intelligent scoring system using heart rate variability, 12-lead electrocardiogram (ECG), and vital signs where a hybrid sampling-based ensemble learning strategy is proposed to handle data imbalance. The experiments were conducted on a dataset consisting of 564 chest pain patients recruited at the ED of a tertiary hospital. The proposed ensemble-based scoring system was compared with established scoring methods such as the modified early warning score and the thrombolysis in myocardial infarction score, and showed its effectiveness in predicting acute cardiac complications within 72 h in terms of the receiver operation characteristic analysis.

In this paper, a new method based on estimation and line-of-sight (LOS) detection of individual time-of-arrival (TOA) measurement and Kalman filter (KF) is proposed to track a moving target in mixed line-of-sight and non-line-of-sight (LOS/NLOS) environments. In the proposed tracking algorithm, TOA measurements collected by multiple stationary sensors in a wireless sensor network are used. First, a pseudo-measured position is calculated by choosing the point along the circle defined by a given TOA measurement which has the shortest distance to the predicted position of a moving target. The pseudo-measured position is shown to be an approximately unbiased estimate of the true position of the target. Second, each pseudo-measured position calculated is passed to a detector to be identified as either LOS or NLOS. The average of all the selected LOS pseudo-measured positions is then used as a new pseudo-measurement for the KF to track the moving target. Unlike all the existing target tracking algorithms in mixed LOS/NLOS environments, the proposed tracking algorithm is able to perform target tracking even with just one LOS TOA measurement at a given time instance without prior information of the NLOS noise which may be difficult to obtain in practice. Another advantage of the proposed tracking algorithm is its computational efficiency. Simulation results show that the proposed tracking algorithm performs better than some recent tracking algorithms, particularly in severe mixed LOS/NLOS environments.

In this paper, a novel individual angle-of-arrival (AOA) measurement detection method and extended Kalman filter (EKF) based tracking algorithm is proposed. The detection method is used to detect whether an individual AOA measurement is line-of-sight (LOS) or non-line-of-sight (NLOS). After the measurement detection, the selected LOS AOA measurements are then used into an dynamic EKF to track a moving target in mixed LOS/NLOS environments. Different from some traditional NLOS error detection methods, which determine the estimation result of a set of AOA measurements collected at every time step is LOS or not, the proposed method detects each AOA measurement one by one at one time step. This algorithm makes good use of LOS AOA measurements and greatly improves the tracking accuracy of the EKF in mixed LOS/NLOS environments. Simulations implemented under different NLOS percentage scenarios demonstrates the improvement of the classical EKF with the assistance of the proposed measurement detection method for AOA measurement.

In this paper, we propose a voting based online sequential extreme learning machine (VOS-ELM) for single hidden layer feedforward networks (SLFNs) to perform the online sequential multi-class classification. Utilizing the recent voting based extreme learning machine (V-ELM) and the online sequential extreme learning machine (OS-ELM), the newly developed VOS-ELM is able to classify online sequences by learning data one-by-one or chunk-by-chunk with fixed or varying chunk size and to reach a higher classification accuracy than the original OS-ELM. Simulations on several real world classification datasets show that VOS-ELM outperforms OS-ELM as well as several state-of-the-art online sequential algorithms.

Allpass filters have found many applications in signal processing areas. In this paper, an iterative linear programming (LP) algorithm is first presented to solve the minimax phase error design of allpass filters. Then an iterative weighted minimax (WMM) method is presented to design allpass filters with equi-ripple group delay errors. Finally, the iterative WMM method is applied to the design of allpass-filter-based halfband filters with equiripple constant group delays. Design examples demonstrate that the new iterative LP algorithm is very efficient for the minimax allpass filter design, and the iterative WMM method is very effective for the equiripple constant group delay designs of allpass filters and allpass-filter-based halfband filters.

In estimation theory, it is known that prior knowledge of parameters can improve the Cramér-Rao lower bound (CRLB). In this paper, we study the influence of prior knowledge on the CRLB of the estimates of the parameters that describe the trajectory of a moving object (single molecule). Since the CRLB is obtained from the inverse of the Fisher information matrix, we present a general expression of the Fisher information matrix in terms of the image function, the object trajectory and the prior knowledge matrix. Applying this expression to an object moving linearly in a two-dimensional (2D) plane with two distinct cases of prior knowledge, explicit CRLB expressions are derived. From these expressions, we show that the improvement in the CRLB of the parameter estimates is dependent on which parameters are known.

A simple but effective approach based on gating and robust extended Kalman filtering (EKF) is proposed for target tracking in mixed line-of-sight and non-line-of-sight (LOS/NLOS) environments, using time-of-arrival (TOA) measurements. Utilizing a gating based detection method determined by the statistical properties of the TOA measurements, a confidence region is derived. Most of the LOS TOA measurements can be effectively selected from all mixture measurements using the properly defined confidence region. These selected LOS TOA measurements are then passed on to the dynamic robust EKF to do the target tracking. Simulations have shown the proposed tracking approach has outperformed both the classic rEKF as well as a recent target tracking approach. Moreover, the proposed approach does not require any statistical knowledge of the NLOS error and only assumes that the standard deviation of the LOS noise is known.

This paper is concerned with the problem of robust unbiased H∞filtering for uncertain two-dimensional (2-D) systems described by the Fornasini-Marchesini local state-space second model. The parameter uncertainties are assumed to be norm-bounded in both the state and measurement equations. The concept of robust unbiased filtering is first introduced into uncertain 2-D systems. A necessary and sufficient condition for the existence of robust unbiased 2-D H∞filters is derived based on the rank condition of the given system matrices. A method is then proposed for the design of robust unbiased H∞filters for uncertain 2-D systems using a linear matrix inequality (LMI) technique. The main advantage of the proposed method is that it can be applied to unstable uncertain 2-D systems while existing robust 2-D H∞filtering approaches only work for robust stable uncertain 2-D systems. An illustrative example is also provided and comparison with existing results is made.

In this paper, we consider the theoretical bound of the probability of error in compressed sensing (CS) with the Bayesian approach. In the detection problem, the signal is sparse and is reconstructed from a compressed measurement vector. Utilizing the oracle estimator in CS, we provide a theoretical bound of the probability of error when the noise in CS is white Gaussian noise (WGN). We show that without any additional information in CS, the probability of error obtained using the signal reconstructed by four recovery algorithms: the basis pursuit denoising (BPDN) algorithm, the Dantzig selector, the orthogonal matching pursuit (OMP) method and the compressive sampling matching pursuit (CoSaMP) algorithm is always larger than the derived theoretical bound. Simulation results demonstrate the effectiveness of our result.

This paper proposes an approach to improving the target tracking accuracy in mixed line-of-sight/non-line-of-sight (LOS/NLOS) environments based on individual measurement detection (IMD) and road constraints. Utilizing the IMD algorithm, most LOS time-difference-of-arrival (TDOA) measurements can be correctly selected from mixed LOS/NLOS TDOA measurements. Incorporating the prior knowledge of the road constraint as a pseudo-measurement, the augmented dynamic extended Kalman filter (EKF) is then implemented to track a moving target. Simulation results demonstrate that the road constraint assisted algorithm, in conjunction with the IMD method, performs better than the road constraint assisted algorithm without the IMD method, or tracking algorithm with the IMD method but without the road constraint. The proposed method is especially useful when the number of selected LOS TDOA measurements is less than three.

Traditional risk score prediction is based on vital signs and clinical assessment. In this paper, we present an intelligent scoring system for the prediction of cardiac arrest within 72 h. The patient population is represented by a set of feature vectors, from which risk scores are derived based on geometric distance calculation and support vector machine. Each feature vector is a combination of heart rate variability (HRV) parameters and vital signs. Performance evaluation is conducted on the leave-one-out cross-validation framework, and receiver operating characteristic, sensitivity, specificity, positive predictive value, and negative predictive value are reported. Experimental results reveal that the proposed scoring system not only achieves satisfactory performance on determining the risk of cardiac arrest within 72 h but also has the ability to generate continuous risk scores rather than a simple binary decision by a traditional classifier. Furthermore, the proposed scoring system works well for both balanced and imbalanced datasets, and the combination of HRV parameters and vital signs shows superiority in prediction to using HRV parameters only or vital signs only.

This paper considers the Roesser state-space model realization problem for a multidimensional (n-D) system by using elementary operations. The new elementary operation approach proposed recently by the authors for the the single-input and single-output (SISO) case will be generalized to the multi- input and multi-output (MIMO) case. Specifically, the realization problem is first formulated as an elementary operation problem of some rc-D polynomial matrix. Then, a general constructive realization procedure is established based on a right matrix fraction description of the given transfer matrix, which guarantees a regular realization and can be easily implemented by a computer program. An example is given to illustrate the details and effectiveness of the proposed approach.

This paper is concerned with the problem of the two-channel 2-D IIR filter bank design. Using a systematic optimization approach, which is solved in terms of an LMI, the 2-D IIR synthesis filters are designed such that the systems are alias free and then optimized to achieve approximate perfect reconstruction.

Non-line-of-sight (NLOS) error is one of the most important factors affecting the accuracy of positioning or tracking especially in urban or indoor environments. This paper concentrates on alleviating the influence of the NLOS error. A position detection approach utilizing the circle error probability (CEP) in conjunction with the least square (LS) method and Kaiman Filter (KF) tracking algorithms is proposed. The LS is used to estimate the positions of a moving target from time-of-arrival (TOA) measurements and the KF is used to smooth the tracking trajectory. Simulation results show that this approach can effectively identify line-of-sight (LOS) estimated positions, leading to higher tracking accuracy than that by other tracking methods without using position detection.

The problem of non-fragile H∞filter design for two-dimensional (2-D) discrete systems in Fornasini-Marchesini model with polytopic uncertainties is considered in this paper. The filters to be designed are assumed to be with additive norm-bounded gain variations. By using the technique of imposing a structural restriction on the slack matrix, we present a simple and yet effective method for designing non-fragile H∞filters. Explicit expressions of the non-fragile H∞filters are given in terms of solutions to a linear matrix inequality (LMI). An illustrative example is provided to demonstrate the feasibility and effectiveness of the proposed method.

The translational motion model of H.264/AVC limits its coding efficiency for video sequences that have more complicated motions. Although an affine motion model was proposed before for overcoming the above limitation, existing methods seem to be complicated to implement. In this paper, we investigate a simple coding scheme based on image alignment to improve the quality for video sequences undergoing a special kind of distortion. With a single global affine transformation applied to a video sequence of this kind, the transformed image frame is geometrically more similar to the target frame. When such an alignment image frame is used as the reference frame, residual error is reduced and hence video quality is improved. Experimental results on both a synthetic video sequence and a real world sequence have shown that the proposed scheme outperforms H.264/AVC reference software and is simpler than existing methods employing affine transformation.

This paper aims at effectively detecting the changes in multi-temporal synthetic aperture radar (SAR) images. Since pixel-based method does not sufficiently utilize the correlation between pixels, it usually may only be used in some specific applications. We propose a change detection method that is based upon discrete cosine transform (DCT) classification. It can utilize the correlation between pixels and eliminate unimportant or nuisance forms of changes. It generates a threshold automatically in DCT classification according to the relationship between DCT coefficient and pixel's gray level. In order to eliminate the block effect, we check pixels not only within the target block, but also inside the neighborhood window around the target block. Experimental results show that the proposed method outperforms some recently published pixel-based methods.

This paper presents an improved method for the design of frequency-response masking (FRM) FIR filters with reduced passband group delays. Expressions of the group delay and its gradient for an FRM filter are derived. To directly and effectively control the group delay error, a group delay constraint is formulated and incorporated into the overall optimization. The effectiveness and flexibility of the proposed design method are demonstrated by an example, which shows that significant reduction in group delay error could be achieved at the cost of slight increase in magnitude error.

The open problem of absolutely minimal state-space realization for two-dimensional (2D) digital filters is revisited. A thorough investigation is carried out for the class of first order 2D digital filters. It is found that the canonical form proposed by Kung et al. three decades ago is incomplete, probably due to an incorrect assumption, which is also pointed out for the first time in this paper. Three new canonical forms for absolutely minimal state-space realization are proposed which can realize any 2D causal digital filters of first order. Several examples are illustrated

An important concept in multidimensional (nD) systems and signal processing is that of equivalence of systems, which is closely related to equivalence of multivariate (nD) polynomial matrices. By exploiting recent results on nD polynomial matrix factorizations, this paper presents several new results on the reduction by equivalence of a given nD polynomial matrix to a simpler form. In particular, the new results work for both 2D and nD (n > 2) cases. The paper also discusses equivalence of nD polynomial matrices that are not of normal full-rank, which have received little attention in the past.

Approximately linear phase FIR filters with reduced group delays are desirable in certain applications. Conventional design methods indirectly achieve the desired group delay by approximating the complex frequency response, resulting in large group delay error, especially at the vicinity of the band edge. A new design method based on semidefinite programming is proposed. An explicit expression for the group delay of an FIR filter is given. With some approximations, a linear matrix inequality constraint on group delay is formulated so that approximately linear phase FIR filters with reduced group delays can be designed using the powerful semidefinite programming approach. An illustrative design example is presented.

This paper presents a scalable fast mode decision algorithm to effectively choose the best coding mode among the 7 macroblock (MB) coding modes that could be used in H.264 video coding. The spatial and temporal complexity of the MB is analyzed to determine the order of searching in the mode decision priority queue such that the most probable mode is checked first, followed by the second most probable mode, and so on. This process is terminated as soon as the computed rate-distortion (RD) cost is below a threshold which is dependent on the complexity ratio of the current MB. By adjusting the threshold we can choose a preferred tradeoff between timesaving and quality compromise. Experimental results show that the proposed fast mode decision algorithm can drastically reduce the encoding time up to 50% with negligible loss of coding efficiency.

This paper introduces characterization of over-sampled linear phase perfect reconstruction filter bank as an embedded part of corresponding critically sampled filter bank. We deal with FIR linear phase perfect reconstruction filter banks with real-valued coefficients in which all analysis filters have the same length. Two methods are presented for designing such filter banks with integer oversampling ratio and rational oversampling ratio, respectively. Finally, several design examples are presented to confirm the validity of our approach.

This letter proposes an improved method for designing complex finite impulse response (FIR) filters with reduced group delays using semidefinite programming. A key step in the proposed method is to directly impose a group delay constraint, which is formulated as a linear matrix inequality constraint with some reasonable approximations. The main advantage of the proposed design method is the significant reduction in group delay error at the expense of the slight increase in magnitude error. The effectiveness of the proposed design method is illustrated with an example

This paper aims at improving perceptual quality of encoded video sequences, and proposes a new perceptual bit allocation scheme for H.264. Firstly, a new motion complexity measure is defined to represent the amount of motion contents between two consecutive frames, and is used to estimate the target bit at frame level. Secondly, a segmentation method exploiting the perceptual characteristics of the video contents is presented, and is used to adaptively update the Lagrangian multiplier in the coding mode selection at macroblock level. Thirdly, based on the motion complexity and Lagrangian multiplier updating, a rate control scheme for H.264 is proposed. Experimental results show that our scheme can effectively improve the perceptual video quality as compared with the H.264 adopted rate control algorithm. Moreover, our scheme also achieves an average peak signal-to-noise ratio gain of up to 0.138 dB for the test sequences.

Calculation of the Cramer-Rao lower bound, i.e., the inverse of the Fisher information matrix, for output data sets of a general nonlinear system is a challenging problem and is considered in this letter. It is shown that the Fisher information matrix for a data set generated by a nonlinear system with additive Gaussian measurement noise can be expressed in terms of the outputs of its derivative system that is also a nonlinear system. An example is considered arising from surface plasmon resonance experiments to determine the dynamic parameters of molecular interactions.

The purpose of this paper is to establish a new constructive procedure for realization of 2D systems by Fornasini-Marchesini second (F-MII) local state-space model which can further reduce the order of the resultant realization to about half of that obtained by the procedure given recently by the authors. Nontrivial illustrative examples are given.

An algorithm is presented to calculate a state space realization of a 3D image set. It is based on interpreting the image set as the impulse response of a 3D separable system. As an application it is shown how the approximation steps, including balanced model reduction methods, in the algorithm can be used to suppress noise in 3D image sets. The approach was motivated by a practical problem in the analysis of 3D fluorescent microscopy image data of fluorescently labelled cells.

A new approach to robust adaptive beamforming for wideband array signals is proposed. General steering vector errors, such as direction-of-arrival mismatch and array positional error, are modeled by "time-delay errors" and compensated for by self-adjusted interpolation filtering. The proposed method effectively overcomes the target-signal cancellation problem without suffering from loss in the degree of freedom for interference rejection, as verified by simulations.

The paper shows how the Fisher information matrix of a given two-dimensional (2D) data set can be expressed using the matrices that determine the 2D system that generates the data set. For uniformly sampled data, it is shown how the Fisher information matrix can be expressed through the solutions of Lyapunov equations. The novel techniques are demonstrated with an example arising from nuclear magnetic resonance spectroscopy.

This paper is a tutorial and review paper on the Grobner bases method and some of its applications in control and systems. Although all the results presented in the paper are available in the literature, we give an elementary treatment here, in the hope that it will further bring awareness of and stimulate interest in Grobner bases among researchers in control and systems. We give a brief tutorial on Euclidean Division Algorithm and Grobner bases, with emphasis on solutions to two important mathematical problems which are closely related to the stabilizability and stabilization of linear Multidimensional systems. Applications of Grobner bases to other areas of control and systems are also briefly mentioned.

Available conventional super resolution SAR imaging methods are based on complex images obtained by SAR imaging algorithms, and form super resolution SAR images via 2-D spectral estimation methods. A super resolution SAR imaging method combined with the chirp scaling algorithm (CSA) is presented in this paper. The new method forms a super resolution image by performing the range and the azimuth processing independently. This method can reduce the effects of phase errors on super resolution processing. The method is suitable for practical application since it does not require heavy computation. Experimental results show the effectiveness of the presented method.

A general method is proposed to factor a discrete W transform (DWT) into lifting steps and additions. Then, based on the relationships among various types of discrete sinusoidal transforms, other types of transforms such as the discrete Fourier transform (DFT) and discrete cosine transform (DCT) are factored into lifting steps and additions. After approximating the lifting matrices, we get various types of new integer discrete transforms such as IntDWT, IntDFT and IntDCT which are floating-point multiplication free. Transforms which map integer to integer are also proposed. Fast algorithms are given for the new transforms and their computational complexities are analyzed. Based on a polynomial transform and an index mapping, multi-dimensional integer transforms are presented with especially low computational complexity.

This paper describes a robust speech detection algorithm that can operate reliably in a microphone array teleconferencing system. High performance in a nonstationary noisy environment is achieved by combining the following techniques: (1) noise suppression by spectral subtraction, (2) silence detection by adaptive noise threshold and (3) non-stationary noise detection based on the availability of pitch signal. This algorithm can prevent the microphone-array-based speaker tracking system from being misguided by noises commonly present in a conference room. Real world experiments show that this algorithm performs very well and has the potential for practical applications.

This paper presents a new coding scheme for arbitrarily shaped image segments using a boundary block grouping and merging (BBGM) technique and a modified shape-adaptive discrete cosine transform (SA-DCT). The BBGM technique merges pairs or groups of neighboring blocks satisfying some properties into larger single blocks. The modified SA-DCT is then applied to these blocks as well as to other boundary blocks. The number of transformed coefficients is the same as the number of foreground pixels in the boundary segment. Simulation results show that the proposed algorithm performs better than the conventional SA-DCT first suggested in 1995 and the pseudo-orthonormal SA-DCT published in 1998.

In this paper, the problem of time-varying parametric autoregressive (AR) model identification by wavelets is discussed. Firstly, we derive multiresolution least squares (MLS) algorithm Gaussian time-varying AR model identification employing wavelet operator matrix representation. This method can optimally balance between the over-fitted solution and the poorly represented estimation. Utilizing multiresolution analysis techniques, the smooth trends and the rapidly changing components of time-varying AR model parameters can both be estimated accurately. Then, the proposed MLS algorithm is combined with the total least squares algorithm for noisy time-varying AR model identification. Simulation results verify the effectiveness of our algorithms.

This paper presents an overview of one of the nD polynomial matrix factorization problems, namely the factorization of a given nD polynomial matrix A(z) into A(z)=A/sub 1/(z)A/sub 2/(z) such that A/sub 1/(z) and A/sub 2/(z) are both nD polynomial matrices with certain special properties. The paper covers applications of both "classical" techniques as well as the Grobner bases to this factorization problem for the 2D and nD (n>2) cases. The difference between the 2D and the nD cases is pointed out. Besides reviewing the existing results, a new result and some open problems are also presented.

Experimental results on the study of helicopter acoustic signals with nonstationary additive noises using cyclostationarity are presented. It is shown that helicopter signals are closely related to first-order periodicity rather than to second- or higher order periodicity, and hence it is necessary to discuss the feasibility of detecting first-order periodicity using cyclostationarity. Working on real helicopter data, we then show that some advantages of using cyclostationarity for helicopter signal detection are improvement of (cyclic) frequency resolution and enhancement of the probability of detection. For helicopter signals with time-variant Doppler shift, a cyclic frequency smoothing method is proposed. This method does not have a counterpart in communication applications, and should be useful for acoustic signal processing.