Automated essay scoring (AES) is a challenging task in cross-prompt settings due to the diversity of scoring criteria. While previous studies have focused on the output of large language models (LLMs) to improve scoring accuracy, we believe activations from intermediate layers may also provide valuable information. To explore this possibility, we evaluated the discriminative power of LLMs’ activations in cross-prompt essay scoring task. Specifically, we used activation to fit probes and further analyzed the effects of different models and input content of LLMs on this discriminative power. By computing the directions of essays across various trait dimensions under different prompts, we analyzed the variation in evaluation perspectives of large language models concerning essay types and traits. Results show that the activations possess strong discriminative power in evaluating essay quality and that LLMs can adapt their evaluation perspectives to different traits and essay types, effectively handling the diversity of scoring criteria in cross-prompt settings.

Fast Adversarial Training (FAT) is known for its efficiency but is often constrained by the limited quality of single-step adversarial examples (AEs), which weakens robust learning signals and leads to unstable optimization. Because of its single-step nature, many generated AEs fail to cross decision boundaries effectively, resulting in diluted robustness improvements. We identify that this limitation stems from a fundamental mismatch: single-step perturbations ignore the semantic vulnerability manifold—low-dimensional subspaces where decision boundaries are thinnest due to class semantic relationships (CSRs). While multi-step attacks implicitly navigate this manifold, single-step methods generate isotropic perturbations that rarely align with these fragile directions. Building on this insight, we propose Semantic-Guided Fast Adversarial Training (SG-FAT), a unified framework that steers single-step attacks along CSR subspaces through three synergistic components operating under a common principle: constrain perturbations to semantic vulnerability directions. SG-FAT improves the quality of single-step adversarial examples and stabilizes training while maintaining computational efficiency. Compared with existing fast AT methods, SG-FAT consistently achieves higher robustness across CIFAR-10, CIFAR-100, and Tiny-ImageNet. Under PGD-10 attacks (ϵ = 8/255), SG-FAT improves robust accuracy by +2.38%, +1.49%, and +1.09%, respectively. Our code and training logs are available at https://github.com/SSonnyboy/SG-FAT.

Low-light image enhancement aims to improve contrast and detail representation under insufficient illumination. However, existing methods primarily rely on deepening or widening convolutional layers, while neglecting image prior information, often resulting in detail loss and visual distortion. Moreover, convolutional neural networks (CNNs) struggle to capture long-range dependencies. To address these limitations, we propose a Gradient-Aware Transformer (GAformer), which integrates gradient-aware convolutions with Transformer-based global modelling. By leveraging gradient priors for enhanced local structural representation and exploiting the global interaction capability of Transformers, GAformer achieves more comprehensive and stable enhancement. Specifically, Gradient-Aware Kernels (GAK) are introduced to optimise edge feature extraction, followed by an Illumination Map-Guided Attention (IGA) mechanism that selectively enhances low-illumination regions. Furthermore, a Frequency-Modulated Calibration (FMC) module facilitates interaction between low- and high-frequency components for progressive guided recovery. Experimental results on multiple benchmark datasets demonstrate that GAformer outperforms state-of-the-art methods in both quantitative evaluation and visual quality.

Abstract Homomorphic signatures have important potential in cloud computing and data privacy protection, but there are still problems such as low signature efficiency, high overhead, and difficulty in instantiation. To solve these problems, an efficient leveled fully homomorphic signature scheme LFHSS with shortened signature values is constructed. The scheme is based on the GPV framework and the RSIS problem on the NTRU lattice. By using the Fast Fourier Sampling algorithm, it achieves efficient signatures with smaller size. It introduces a homomorphic trapdoor function and designs three basic evaluations: homomorphic addition, multiplication, and scalar multiplication. These operations enable the LFHSS scheme to support homomorphic evaluations of functions consisting of addition, multiplication, and scalar multiplication within a certain circuit depth. Additionally, it is proven to be strongly unforgeable, and the scheme is implemented in software with correctness testing and performance analysis conducted. The experimental results show that the security level of LFHSS is 1.14 times higher than that of BCFL23, and the signature generation speed is 300+ times faster, the homomorphic evaluation speed is 7.8 times faster, and the verification speed is 48K times faster than that of BCFL23. The signature length of LFHSS is only 4.86% of BCFL23. The work in this paper is of great significance to the design and application of homomorphic signatures.

Federated Learning (FL) systems are increasingly deployed across multiple clients to efficiently train a shared model over local data, thereby effectively addressing data silos and reducing communication. However, FL systems are known to be susceptible to model poisoning attacks by malicious clients, who aim at deteriorating the global model accuracy through sending corrupted updates to the central server. Meanwhile, the local accuracy discrepancy among clients, called as performance fairness, could also be exacerbated, which is one of the major concerns of trustworthy FL systems. This paper proposes a novel attack framework called Durable Model Poisoning Attack (DMPA), targeting both fairness and robustness of efficient FL systems. To implement DMPA, we design the over-unlearning strategy, enabling the adversary to generate poisoned updates to compromise partial clients' performance. Furthermore, we develop a dual projection mechanism to improve the durability of model poisoning attacks. Extensive experiments demonstrate that DMPA is powerful and effective even against robust aggregation rules. Particularly, DMPA achieves average$7.6\times$higher reduction of accuracy while decreasing the performance fairness by$3.0\times$compared with baselines. The experiments also indicated that DMPA extends the durability of attack impacts over baselines by$8.5\times$. In addition, experiments in efficient FL systems disclose their vulnerability.

Achieving counterfactual fairness is a critical objective in advancing fairness research within machine learning. Studies have shown that machine learning models often inherit biases from their training data, leading to unfair decision-making. Fair data generation methods aim to mitigate these biases, ensuring that predictors trained on such data uphold fairness. However, in the context of counterfactual fairness, existing methods for generating fair data are often limited in their applicability and lead to significant performance losses in downstream predictors. To address these issues, this paper proposes a new algorithm for generating counterfactually fair data, allowing predictors trained on this generated data to adhere to counterfactual fairness. We propose a new metric, Intervention-Driven Correlation (IDC), to evaluate the counterfactual fairness of generative models. IDC assesses fairness by applying random interventions to samples and measuring the statistical correlation between the degree of intervention and the outcome of interest. This metric is applicable to both discrete and continuous sensitive attributes and labels. Furthermore, our studies reveal a critical insight: counterfactually fair data does not always guarantee counterfactually fair predictors when deployed in real-world scenarios. We identify the root causes of this issue and propose a robust solution. To bridge this gap, we propose the IDC-Reduction method, which ensures the fairness of downstream predictors by generating counterfactually fair data. Experimentally, our method outperforms existing approaches and achieves counterfactual fairness regardless of the type of downstream predictors.

Computing power has evolved into a foundational and indispensable resource in the area of deep learning, particularly in tasks such as Face Recognition (FR) model training on large-scale datasets, where multiple GPUs are often a necessity. Recognizing this challenge, some FR methods have started exploring ways to compress the fully-connected layer in FR models. Unlike other approaches, our observations reveal that without prompt scheduling of the learning rate (LR) during FR model training, the loss curve tends to exhibit numerous stationary subsequences. To address this issue, we introduce a novel LR scheduler leveraging Exponential Moving Average (EMA) and Haar Convolutional Kernel (HCK) to eliminate stationary subsequences, resulting in a significant reduction in converging time. However, the proposed scheduler incurs a considerable computational overhead due to its time complexity. To overcome this limitation, we propose FastFace, a fast-converging scheduler with negligible time complexity, i.e.O(1) per iteration, during training. In practice, FastFace is able to accelerate FR model training to a quarter of its original time without sacrificing more than 1% accuracy, making large-scale FR training feasible even with just one single GPU in terms of both time and space complexity. Extensive experiments validate the efficiency and effectiveness of FastFace. The code is publicly available at: https://github.com/amoonfana/FastFace.

In the Internet of Medical Things (IoMT), de novo peptide sequencing prediction is one of the most important techniques for the fields of disease prediction, diagnosis, and treatment. Recently, deep-learning-based peptide sequencing prediction has been a new trend. However, most popular deep learning models for peptide sequencing prediction suffer from poor interpretability and poor ability to capture long-range dependencies. To solve these issues, we propose a model named SeqNovo, which has the encoding-decoding structure of sequence to sequence (Seq2Seq), the highly nonlinear properties of multilayer perceptron (MLP), and the ability of the attention mechanism to capture long-range dependencies. SeqNovo use MLP to improve the feature extraction and utilize the attention mechanism to discover key information. A series of experiments have been conducted to show that the SeqNovo is superior to the Seq2Seq benchmark model, DeepNovo. SeqNovo improves both the accuracy and interpretability of the predictions, which will be expected to support more related research.

The adversarial vulnerability of convolutional neural networks (CNNs) refers to the performance degradation of CNNs under adversarial attacks, leading to incorrect decisions. However, the causes of adversarial vulnerability in CNNs remain unknown. To address this issue, we propose a unique cross-scale analytical approach from a statistical physics perspective. It reveals that the huge amount of nonlinear effects inherent in CNNs is the fundamental cause for the formation and evolution of system vulnerability. Vulnerability is spontaneously formed on the macroscopic level after the symmetry of the system is broken through the nonlinear interaction between microscopic state order parameters. We develop a cascade failure algorithm, visualizing how micro perturbations on neurons' activation can cascade and influence macro decision paths. Our empirical results demonstrate the interplay between microlevel activation maps and macrolevel decision-making and provide a statistical physics perspective to understand the causality behind CNN vulnerability. Our work will help subsequent research to improve the adversarial robustness of CNNs.

Vehicle re-identification is a crucial task that matches images of the same vehicle across different camera viewpoints. Many previous attention-based studies have approached this problem by exploring the regions of interest in vehicles. However, the generated attention in these models is susceptible to noisy data, as they are unable to provide powerful supervision to distinguish biased and unbiased clues during the attention learning process. To address the problems mentioned above, we aim to design a robust vehicle re-ID network that utilizes the causal effect to effectively transfer attention from biased to unbiased clues. In this paper, we propose an unbiased feature-enhanced network (UnbiasNet), which consists of an unbiased feature-aware block (UFAB) and a novel causal effect-based joint constraint (CEC). In particular, we propose an unbiased feature-aware block as an attention module to extract rich and discriminative information. We conduct a counterfactual intervention on our attention module to generate biased feature representations. Moreover, we propose a novel causal effect-based joint constraint that consists of original prediction constraint and total indirect effect constraint. The original prediction constraint ensures that unbiased feature-aware block converges correctly. The total indirect effect constraint utilizes the generated biased features as supervisory information to motivate unbiased feature-aware block to explore a greater number of unbiased features during the training process. Our approach had an inference time of 1.39 ms per image, which introduces only a few parameters during the training phase and none during the testing phase. We carry out comprehensive experiments to illustrate the effectiveness of the UnbiasNet on three challenging datasets.

Deep neural networks (DNNs) have immense potential for precise clinical decision-making in the field of biomedical imaging. However, accessing high-quality data is crucial for ensuring the high-performance of DNNs. Obtaining medical imaging data is often challenging in terms of both quantity and quality. To address these issues, we propose a score-based counterfactual generation (SCG) framework to create counterfactual images from latent space, to compensate for scarcity and imbalance of data. In addition, some uncertainties in external physical factors may introduce unnatural features and further affect the estimation of the true data distribution. Therefore, we integrated a learnable FuzzyBlock into the classifier of the proposed framework to manage these uncertainties. The proposed SCG framework can be applied to both classification and lesion localization tasks. The experimental results revealed a remarkable performance boost in classification tasks, achieving an average performance enhancement of 3-5% compared to previous state-of-the-art (SOTA) methods in interpretable lesion localization.

Information granule is a classic mathematical paradigm in the field of data mining. Existing research focuses on improving granule quality to optimize models. However, in these studies, they do not consider that the fairness of the granular model will affect the performance of the granule, especially in some severe social issues (Law, Finance, Education, and more), often due to the participation of sensitive features, the application results of the granule (such as classification) cause population bias. Thus, we construct an adaptive probability replacement resampling model (APR) in Pre-Processors to reduce the bias of sensitive features on granular results. Then, in In-Processors, we add fairness optimizations (FO) to improve the fairness of information granules. Finally, we propose a fair information granule model based on adaptive probability replacement resampling called APRG(APR+FO). We select three loan datasets to verify the feasibility of the granular model according to the current sensitive lending issues in the financial field. The experimental results show that compared with the existing granular models, our proposed method dramatically improves the fairness of the constructed granular models, especially the APRG model, which has an average increase of 46.6% on Demographic Parity (DP) and 77.9% on Equalized Odds (EO). Compared with other existing granular models, the granule quality of APRG is increased by 38.88% on average; in terms of classification accuracy, the average decrease is only 2.03%.

Top-k dominance (TKD) query is an extended query method of skyline query and top-k query, which reveals the top-k dominant individuals in an incomplete dataset by analyzing the dominance relationships between individuals and is a common decision tool in intelligent recommendation applications. This research proposes two parallel query algorithms based on Spark computing engine to address the shortcomings of the parallel top-k-dominated query algorithms for dynamic incomplete datasets. The designed model achieves good performance in terms of runtime performance compared to previous studies.

In recent years, voice-interaction-based control systems have attracted considerable attention for industrial control systems implementing Industrial Internet of Things (IIoT) technologies. The development of automated semantic understanding relates to the industrial Internet equipment used to realize remote voice control as well as to its intelligent management and control. In these emerging voice-interaction-enabled industrial central control systems, sorting technologies are considered critical. For complex user questions, the level of satisfaction regarding the answers given by such systems tends to be low. Driven by these challenges and opportunities, the optimization of conventional retrieval-based question answering through deep learning methods has become popular. In this study, we propose three deep semantic sorting models based on deep learning, including a multilayer convolutional matching sorting model for single documents and two interactive pairwise bidirectional encoder representations from transformers (BERT) sorting models for document pairs. Two main network architectures are proposed to model document pairs, named Pairwise-Twin-BERT and Pairwise-Triple-BERT. Experimental results indicate that proposed models performed better than state-of-the-art methods based on text matching in a candidate document sorting task.

Artificial intelligence-driven automation has gradually become the technical trend of the new automation era. At present, many artificial intelligence technologies have been applied to improve the intelligence level in the field of automation. Among them, convolutional neural network (CNN) technology is one of the most representative, which is used in the detection of defective products in industrial automation, robot human tracking has been widely used in the field of machine vision driven automation. However, the high dependence of the current neural network application leads to the potential failure of the defective product detection system. In this article, we model the learning and decision-making process of CNN with a statistical physical percolation model. Based on the differentiation degree and vulnerability of percolation, we propose the concept of CNN differentiation degree and summarize the empirical formula to quantify it. The relationship between the differentiation degree and vulnerability is analyzed from both adversarial attack and adversarial training perspectives to explain the decision-making mechanism of CNN and classification reliability. The physical model can approach the essence of things and finally guide the reliable CNN for industrial automation.

In this article, we propose and analyze a group of adversarial backdoor attack methods on neural-architecture-search (NAS) enabled edge AI systems in industrial Internet of Things (IIoT) domain. NAS is a new popular way to generate scale-adaptive deep neural networks which can meet the respective requirements of cloud, edge, and terminal AI computing in IIoT domain. However, since most users in NAS-enabled edge side are not the generators of AI models, the deployed edge AI models may have some vulnerabilities such as backdoors. These might pose serious security issues in IIoT. We propose some effective policies to attack such edge AI systems and provide advice about how to defend them. The most significant attack through third-party pretrained NAS in IIoT may occur by backdoor attacks while the third party might introduce vulnerability in the training dataset. The article designs backdoor attack processes to NAS-enabled edge devices to identify NAS’s vulnerability to adversarial trojaning attacks and interpret the backdoor attacks. It shows that the existence of high impact nodes greatly weakens the robustness of the network. A malicious attacker can quickly paralyze the network by only selecting a few high impact nodes. Finally, it provides advice and possible solution on defending the adversarial backdoor attacks to NAS.

Edge AI technologies have been used for many Intelligent Transportation Systems, such as road traffic monitor systems. Neural Architecture Search (NAS) is a typcial way to search high-performance models for edge devices with limited computing resources. However, NAS is also vulnerable to adversarial attacks. In this paper, A One-Shot NAS is employed to realize derivative models with different scales. In order to study the relation between adversarial robustness and model scales, a graph-based method is designed to select best sub models generated from One-Shot NAS. Besides, an evaluation method is proposed to assess robustness of deep learning models under various scales of models. Experimental results shows an interesting phenomenon about the correlations between network sizes and model robustness, reducing model parameters will increase model robustness under maximum adversarial attacks, while, increasing model paremters will increase model robustness under minimum adversarial attacks. The phenomenon is analyzed, that is able to help understand the adversarial robustness of models with different scales for edge AI transportation systems.

In recent years, Internet of Things (IoT)-enabled health monitoring wearable devices have become a trend in healthcare systems, regularly collecting vital sign data from patients and uploading them to the cloud. Through on-demand search queries, data are shared with third-party healthcare service providers to monitor patients' health status and provide timely diagnoses. To ensure privacy and security, patient health data should be encrypted before being uploaded to the cloud. The cloud can give search encryption services. However, current searchable encryption (SE) technologies still have problems with forward privacy security and verifiability. This article proposes an IoT-cloud-enabled healthcare data system incorporating a SE method with forward privacy and verifiability. By designing a trapdoor permutation function, we render the resulting output indistinguishable from meaningless random data to the adversary. Thus, the adversary cannot judge the relationship between a newly inserted record and a past search token, and therefore, the system realizes forward privacy or forward secrecy. We propose a multikeyword search verification mechanism based on a pseudo-random function. Our approach solves verifying the correctness of search results in the top-k search scenario with partial search results.Aformal security analysis proves that our scheme achieves forward privacy preservation, which can help guarantee healthcare data privacy. Additionally, a performance evaluation shows that our method is efficient and effective, providing an information security system to preserve patient privacy in IoT-enabled healthcare systems.

With the construction of intelligent transportation, big data with heterogeneous, multi-source and massive characteristics has become an important carrier of cooperative intelligent transportation systems (C-ITS) and plays an important role. Big data in C-ITS can break through the restrictions between regions and entities and then learning cooperatively by sharing data. In addition, the combined efficiency and information integration advantages of big data are conducive to the construction of a comprehensive and three-dimensional traffic information system and can enhance traffic prediction. However, such substantial sensitive data, mainly on the cloud infrastructure, exposes several vulnerabilities like data leakages and privacy breaks, especially when data is shared for cooperative learning purposes. To address this, this paper proposes a forward privacy-preserving scheme, named AFFIRM, for multi-party encrypted sample alignment adopting cooperative learning in C-ITS. By introducing the searchable encryption method, we realize the sample alignment of cooperative learning in the multi-party encrypted data space. AFFIRM ensures encrypted sample alignment under the condition of forward privacy security. We have formally proved that the proposed scheme satisfies both forward security and validity. We have assessed AFFIRM by validating the potential threat of malicious tampering by privacy attackers and malicious personnel search for the aligned sample data and verify it. Finally, we numerically tested and compared AFFIRM against the corresponding ones of some state-of-the-art schemes under various record sizes, servers and processing.

Deep learning-based auto-driving systems are vulnerable to adversarial examples attacks which may result in wrong decision making and accidents. An adversarial example can fool the well trained neural networks by adding barely imperceptible perturbations to clean data. In this paper, we explore the mechanism of adversarial examples and adversarial robustness from the perspective of statistical mechanics, and propose an statistical mechanics-based interpretation model of adversarial robustness. The state transition caused by adversarial training based on the theory of fluctuation dissipation disequilibrium in statistical mechanics is formally constructed. Besides, we fully study the adversarial example attacks and training process on system robustness, including the influence of different training processes on network robustness. Our work is helpful to understand and explain the adversarial examples problems and improve the robustness of deep learning-based auto-driving systems.

6G technology enables artificial intelligence (AI)-based massive IoT to manage network resources and data with ultra high speed, responsive network, and wide coverage. However, many AI-enabled Internet-of-Things (AIoT) systems are vulnerable to adversarial example attacks. Therefore, designing robust deep learning models that can be deployed on resource-constrained devices has become an important research topic in the field of 6G-enabled AIoT. In this article, we propose a method for automatically searching for robust and efficient neural network structures for AIoT systems. By introducing a skip connection structure, a feature map with reduced front-end influence can be used for calculations during the classification process. Additionally, a novel type of densely connected search space is proposed. By relaxing this space, it is possible to search for network structures efficiently. In addition, combined with adversarial training and model delay constraints, we propose a multiobjective gradient optimization method to realize the automatic searching of network structures. Experimental results demonstrate that our method is effective for AIoT systems and superior to state-of-the-art neural architecture search algorithms.

At present, specific voice control has gradually become an important means for 5G-Internet-of-Things-aided industrial control systems, such as controlling the operation and adjustment of industrial Internet of Things equipment through telephone voice of the controller. However, the security of specific voice control system needs to be improved, because the voice cloning technology based on transfer learning can easily simulate the voice of the controller, which may lead to industrial accidents and other potential security risks. Therefore, this article mainly aims to study and understand the principle of voice cloning attack technology, putting forward a voice clone attack method, in order to prepare for the construction of a specific voice recognition system in the future. At present, the key technology of voice cloning attack is how to solve the problem that the target speaker's personalized speech with high quality cannot be synthesized under small samples. In fact, voice cloning is a very challenging problem because speech is more difficult to be represented in the hidden space of the model. We propose a transductive voice transfer learning method to learn the predictive function from the source domain and fine-tune in the target domain adaptively. The target learning task and the source learning task are both synthesizing speech signals from the given audio, while the datasets of both domains are different. By adding different penalty values to each instances and minimizing the expected risk, an optimal precise model can be learned. In addition, an evaluation method to verify the audio similarity of the target speaker was given to show the similarity between the synthesized audio and the original audio. Many details of the experimental results show that our method can effectively synthesize the speech of the target speaker with small samples.

Opportunistic mobile social network (OMSN) enables users to form an instant social network for information sharing (e.g., people watching the same soccer game can share their instant comments). OMSN is ad hoc in nature, thus relies on the cooperation of members regarding message transmission. However, some uncooperative or malicious behavior from abnormal members may reduce network performance, even damage the entire network. Currently, there does not exist effective mechanisms to detect selfish and malicious nodes. To tackle this problem, we propose a dynamic trust framework to facilitate a node to derive a trust value of another node based on the behavior of the latter. The novelty of our framework includes the following: 1) we design a new metric for a trust value of a node and 2) we propose a “two-hop feedback method” that requires intermediate nodes in a forwarding path to generate ACK messages to verify a node's honesty if they are two hops away. In most existing trust models, final ACK messages are considered as critical factors. In OMSN, nodes are not fully connected and final ACK messages cannot be reliably received. In order to avoid the problem that few final ACK messages can be received, we propose a “two-hop feedback method.” Simulation results show that our approach is able to detect a majority of abnormal nodes including malicious nodes, selfish nodes, and those nodes launching conspiracy attacks. Thus, the entire network efficiency can be improved without negative impact of abnormal nodes. Besides, our trust framework can be easily applied to the current popular routing protocols of opportunistic networks.

Recently, researchers found a new type of attacks, called time synchronization attack (TS attack), in cyber-physical systems. Instead of modifying the measurements from the system, this attack only changes the time stamps of the measurements. Studies show that these attacks are realistic and practical. However, existing detection techniques, e.g. bad data detection (BDD) and machine learning methods, may not be able to catch these attacks. In this paper, we develop a "first difference aware" machine learning (FDML) classifier to detect this attack. The key concept behind our classifier is to use the feature of "first difference", borrowed from economics and statistics. Simulations on IEEE 14-bus system with real data from NYISO have shown that our FDML classifier can effectively detect both TS attacks and other cyber attacks.

Dramatic growth of e-book sales revenue in recent years makes book recommendations essential to readers. Traditional bag-of-words models have difficulty of capturing the spatial information of terms over books. In this paper, a three-layer tree structure is used for representing each book. A framework, Tree2Vector, is designed for transforming tree-based book data into vectorial space. First, in order to characterize the global discriminative information of child nodes conveyed at the same level of all the trees, a clustering technique is used for assigning child nodes into different clusters, which are adopted for formulating the components of a vector. Furthermore, a locality reconstruction (LR) method is designed to model the reconstruction process, where each parent node is supposed to be reconstructed by its child nodes. The derived reconstruction coefficients are used for locally weighting the components of the vector. The process is repeated level-by-level until a vectorial representation is accomplished for a book tree. Our method is examined in content-based book recommendation. Experimental results exhibit the effectiveness of our framework.

In this paper, we present a new method for human object-level video advertising. A framework that aims to embed content-relevant ads within a video stream is investigated in this context. In particular, to support content-relevant advertising, we employ the discriminatively trained part based model to detect human objects in a video and then select the ads that are related to the detected human objects. For human clothing advertising, we design a deep Convolutional Neural Network (CNN) using face features to recognize human genders in a video stream. Human parts alignment is then implemented to extract human part features that are used for clothes retrieval. Our novel framework is examined in various types of videos. Experimental results demonstrate the effectiveness of the proposed method for human object-level video advertising.

Semi-supervised clustering is one of the most active research area in machine learning and pattern recognition, which can improve the performance of unsupervised clustering efficiently. This paper focuses on exploiting both the label information of a few labeled samples and the spatial distribution information of large amount of unlabeled samples. We proposed a new semi-supervised clustering method, named Bottleneck Distance based Semi-supervised Clustering (BDSC), which is based on the idea of label propagation and can perform clustering with no parameters. BDSC works by firstly obtaining small amount of labeled samples for each class. Then, a minimum spanning tree is constructed from both labeled and unlabeled samples, where the distances between an unlabeled sample and labeled samples are computed to get the bottleneck distance for each unlabeled sample. Finally, labels are propagated by comparing the bottleneck distances. Experimental results demonstrate that the proposed technique outperforms classical clustering algorithms with respect to the precision and the capability of recognizing nonspherical-shaped clusters.

Currently how to make the hearing aids more and more intelligent has attracted our interests. In order to realize adaptive amplification strategy to improve the audibility, context-aware hearing is crucial. In context-aware hearing, a big difficulty to be solved is context sensing since it is not applicable to implant multiple sensors in the limited space of hearing devices. Therefore, we propose a new context recognition scheme which adopt smart phone to collect sensing data and infer a scene to adapt level of hearing aid. A context recognition framework with a context reasoning model for scene recognition and activity recognition are given. Once scene and activity are confirmed, the smart phone would send a command to hearing aid to actuate the amplification process by Bluetooth. Since smart phones are carried by people in normal life, employing smart phone to sense context data and inference scene is quite a reasonable way to improve the adaptiveness of hearing aids for most people without any other extra devices or sensors. Our contribution in this paper is that let smart phones stay in pockets or bags as exact normal life, the scene can still be inferred to conduct hearing aids. Besides, we conduct some experiments, the results are encouraging and time cost is acceptable.