The pervasive issue of bias in AI presents a significant challenge to painting classification, and is getting more serious as these systems become increasingly integrated into tasks like art curation and restoration. Biases, often arising from imbalanced datasets where certain artistic styles dominate, compromise the fairness and accuracy of model predictions, i.e., classifiers are less accurate on rarely seen paintings. While prior research has made strides in improving classification performance, it has largely overlooked the critical need to address these underlying biases, that is, when dealing with out-of-distribution (OOD) data. Our insight highlights the necessity of a more robust approach to bias mitigation in AI models for art classification on biased training data. We propose a novel OOD-informed model bias adaptive sampling method called BOOST (Bias-Oriented OOD Sampling and Tuning). It addresses these challenges by dynamically adjusting temperature scaling and sampling probabilities, thereby promoting a more equitable representation of all classes. We evaluate our proposed approach to the KaoKore and PACS datasets, focusing on the model’s ability to reduce class-wise bias. We further propose a new metric, Same-Dataset OOD Detection Score (SODC), designed to assess class-wise separation and per-class bias reduction. Our method demonstrates the ability to balance high performance with fairness, making it a robust solution for unbiasing AI models in the art domain.

• Unified text- and image-guided 3D style transfer with consistent views across a scene, while preserving fine details. • Flow-based feature matching aligns text and image cues for faithful style. • Parallel frequency branches balance structure (low frequency) and detail (high frequency). • Cross-dimension and texture losses improve color and texture fidelity.

• Introduces A 2 D 2 C: attention-driven dynamic convolution for local adaptation. • Uses multi-point random sampling to route and fuse k base kernels efficiently. • Presents A 2 D 2 C ＋ that fuses kernels once, cutting redundancy and MAdds at parity. • Shows consistent gains on ImageNet, CIFAR-100 and COCO with statistical reports.

This study introduces a novel electrocardiogram (ECG) arrhythmia classification model, PathFusion-Net, which integrates Rough Path Theory with deep learning technologies. The model combines Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), Path Signatures, and Path Development to extract spatial morphological features from ECG images and multi-order temporal representations from ECG signals. By adopting an inter-patient split paradigm, our approach more closely reflects real-world clinical diagnostic settings compared to intra-patient methods. The model demonstrates state-of-the-art overall classification performance on both the MIT-BIH Arrhythmia Database and a private clinical dataset, achieving 94.7% and 95.1% accuracy, respectively, under the AAMI four-class standard with an inter-patient split paradigm. On the MIT-BIH dataset, the proposed method attains competitive precision and recall across multiple arrhythmia types, including 95.2% /87.9% for ventricular ectopic beats (V) and 75.7% /92.3% for supraventricular ectopic beats (S), indicating balanced performance across clinically diverse categories. This research highlights the potential of Rough Path Theory in time-series analysis and offers a novel deep learning framework for automated early detection and monitoring of ECG arrhythmias. The code used in this study is available at: https://github.com/Rand2AI/PathFusion-Net.

Federated learning is a machine learning paradigm that enables decentralized clients to collaboratively learn a shared model while keeping all the training data local. While considerable research has focused on federated image generation, particularly Generative Adversarial Networks, Variational Autoencoders have received less attention. In this paper, we address the challenges of non-IID (independently and identically distributed) data environments featuring multiple groups of images of different types. Non-IID data distributions can lead to difficulties in maintaining a consistent latent space and can also result in local generators with disparate texture features being blended during aggregation. We thereby introduce FissionVAE that decouples the latent space and constructs decoder branches tailored to individual client groups. This method allows for customized learning that aligns with the unique data distributions of each group. Additionally, we incorporate hierarchical VAEs and demonstrate the use of heterogeneous decoder architectures within FissionVAE. We also explore strategies for setting the latent prior distributions to enhance the decoupling process. To evaluate our approach, we assemble two composite datasets: the first combines MNIST and FashionMNIST; the second comprises RGB datasets of cartoon and human faces, wild animals, marine vessels, and remote sensing images. Our experiments demonstrate that FissionVAE greatly improves generation quality on these datasets compared to baseline federated VAE models.

Abstract Image restoration poses a significant challenge, aiming to accurately recover damaged images by delving into their inherent characteristics. Various models and algorithms have been explored by researchers to address different types of image distortions, including sparse representation, grouped sparse representation, and low‐rank self‐representation. The grouped sparse representation algorithm leverages the prior knowledge of non‐local self‐similarity and imposes sparsity constraints to maintain texture information within images. To further exploit the intrinsic properties of images, this study proposes a novel low‐rank representation‐guided grouped sparse representation image restoration algorithm. This algorithm integrates self‐representation models and trace optimization techniques to effectively preserve the original image structure, thereby enhancing image restoration performance while retaining the original texture and structural information. The proposed method was evaluated on image denoising and deblocking tasks across several datasets, demonstrating promising results.

Spatial convolution is fundamental in constructing deep Convolutional Neural Networks (CNNs) for visual recognition. While dynamic convolution enhances model accuracy by adaptively combining static kernels, it incurs significant computational overhead, limiting its deployment in resource-constrained environments such as federated edge computing. To address this, we propose Fast Multi-Attention Dynamic Convolution (FMDConv), which integrates input attention, temperature-degraded kernel attention, and output attention to optimize the speed-accuracy trade-off. FMDConv achieves a better balance between accuracy and efficiency by selectively enhancing feature extraction with lower complexity. Furthermore, we introduce two novel quantitative metrics, the Inverse Efficiency Score and Rate-Correct Score, to systematically evaluate this trade-off. Extensive experiments on CIFAR-10, CIFAR-100, and ImageNet demonstrate that FMDConv reduces the computational cost by up to 49.8% on ResNet-18 and 42.2% on ResNet-50 compared to prior multi-attention dynamic convolution methods while maintaining competitive accuracy. These advantages make FMDConv highly suitable for real-world, resource-constrained applications. This figure presents an overview of the proposed Fast Multi-Attention Dynamic Convolution (FMDConv) framework, which integrates Input Attention, Temperature-Degraded Kernel Attention, and Output Attention to optimize the speed-accuracy trade-off in convolutional neural networks. The diagram illustrates how these mechanisms enhance feature selection at different stages, significantly reducing computational cost while maintaining competitive accuracy, making FMDConv suitable for resource-constrained applications. • Introduces IES and RCS to quantify speed-accuracy trade-off in CNNs. • Evaluates channel, kernel, and filter attention for effective structures. • Develops a new CNN with kernel attention to enhance efficiency and accuracy. • Demonstrates FMDConv’s advantages via standard benchmark testing.

Labeling large amounts of medical data is travailing, leading to the blooming of few-shot medical image segmentation, which aims to segment the foreground of a query image given a labeled support set. Almost all current models adopt the cosine distance to measure the similarity between prototypes and query features. However, the limitation of the cosine distance is exacerbated by intra-class differences and inter-class imbalances in medical image scenarios, where angle-only evaluation can induce misclassification to under- and over-segmentation. Motivated by this, we propose a High-Confidence Prior Mask-guided Network (HCPMNet), comprising a High-Confidence Mask Generator (HCPMG), a Target Region Mining (TRM) module, and a Prototype-Oriented Expansion Match (POEM) module. Our HCPMNet offers key advantages: 1) HCPMG is the first to combinatively evaluate angle and magnitude similarity, generating high-confidence priori masks that accurately and completely localize target regions. 2) TRM mines and aggregates target class information under the guidance of priori masks. 3) POEM, based on both similarity metrics, correctly matches prototypes with query features. Extensive experiments on three general medical datasets show that our HCPMNet achieves a new SoTA with great superiority.

Abstract Image restoration, as a fundamental research topic of image processing, is to reconstruct the original image from degraded signal using the prior knowledge of image. Group sparse representation (GSR) is powerful for image restoration; it however often leads to undesirable sparse solutions in practice. In order to improve the quality of image restoration based on GSR, the sparsity residual model expects the representation learned from degraded images to be as close as possible to the true representation. In this article, a group residual learning based on low‐rank self‐representation is proposed to automatically estimate the true group sparse representation. It makes full use of the relation among patches and explores the subgroup structures within the same group, which makes the sparse residual model have better interpretation furthermore, results in high‐quality restored images. Extensive experimental results on two typical image restoration tasks (image denoising and deblocking) demonstrate that the proposed algorithm outperforms many other popular or state‐of‐the‐art image restoration methods.

Conventional machine learning methodologies require the centralization of data for model training, which may be infeasible in situations where data sharing limitations are imposed due to concerns such as privacy and gradient protection. The Federated Learning (FL) framework enables the collaborative learning of a shared model without necessitating the centralization or sharing of data among the data proprietors. Nonetheless, in this paper, we demonstrate that the generalization capability of the joint model is suboptimal for Non-Independent and Non-Identically Distributed (Non-IID) data, particularly when employing the Federated Averaging (FedAvg) strategy as a result of the weight divergence phenomenon. Consequently, we present a novel boosting algorithm for FL to address both the generalization and gradient leakage challenges, as well as to facilitate accelerated convergence in gradient-based optimization. Furthermore, we introduce a secure gradient sharing protocol that incorporates Homomorphic Encryption (HE) and Differential Privacy (DP) to safeguard against gradient leakage attacks. Our empirical evaluation demonstrates that the proposed Federated Boosting (FedBoosting) technique yields significant enhancements in both prediction accuracy and computational efficiency in the visual text recognition task on publicly available benchmarks.

Federated Learning (FL) has emerged as a powerful paradigm for training Machine Learning (ML), particularly Deep Learning (DL) models on multiple devices or servers while maintaining data localized at owners’ sites. Without centralizing data, FL holds promise for scenarios where data integrity, privacy and security and are critical. However, this decentralized training process also opens up new avenues for opponents to launch unique attacks, where it has been becoming an urgent need to understand the vulnerabilities and corresponding defense mechanisms from a learning algorithm perspective. This review paper takes a comprehensive look at malicious attacks against FL, categorizing them from new perspectives on attack origins and targets, and providing insights into their methodology and impact. In this survey, we focus on threat models targeting the learning process of FL systems. Based on the source and target of the attack, we categorize existing threat models into four types, Data to Model (D2M), Model to Data (M2D), Model to Model (M2M) and composite attacks. For each attack type, we discuss the defense strategies proposed, highlighting their effectiveness, assumptions and potential areas for improvement. Defense strategies have evolved from using a singular metric to excluding malicious clients, to employing a multifaceted approach examining client models at various phases. In this survey paper, our research indicates that the to-learn data, the learning gradients, and the learned model at different stages all can be manipulated to initiate malicious attacks that range from undermining model performance, reconstructing private local data, and to inserting backdoors. We have also seen these threat are becoming more insidious. While earlier studies typically amplified malicious gradients, recent endeavors subtly alter the least significant weights in local models to bypass defense measures. This literature review provides a holistic understanding of the current FL threat landscape and highlights the importance of developing robust, efficient, and privacy-preserving defenses to ensure the safe and trusted adoption of FL in real-world applications. The categorized bibliography can be found at: https://github.com/Rand2AI/Awesome-Vulnerability-of-Federated-Learning.

With increasing concerns over data privacy and model copyrights, especially in the context of collaborations between AI service providers and data owners, an innovative Sentinel-Guided Zero-Shot Learning (SG-ZSL) paradigm is proposed in this work. SG-ZSL is designed to foster efficient collaboration without the need to exchange models or sensitive data. It consists of a teacher model, a student model and a generator that links both model entities. The teacher model serves as a sentinel on behalf of the data owner, replacing real data, to guide the student model at the AI service provider’s end during training. Considering the disparity of knowledge space between the teacher and student, we introduce two variants of the teacher model: the omniscient and the quasi-omniscient teachers. Under these teachers’ guidance, the student model seeks to match the teacher model’s performance and explores domains that the teacher has not covered. To trade-off between privacy and performance, we further introduce two distinct security-level training protocols: white-box and black-box, enhancing the paradigm’s adaptability. Despite the inherent challenges of real data absence in the SG-ZSL paradigm, it consistently outperforms in ZSL and GZSL tasks, notably in the white-box protocol. Our comprehensive evaluation further attests to its robustness and efficiency across various setups, including stringent black-box training protocol.

Data privacy has become an increasingly important issue in Machine Learning (ML) , where many approaches have been developed to tackle this challenge, e.g., cryptography ( Homomorphic Encryption (HE) , Differential Privacy (DP) ) and collaborative training (Secure Multi-Party Computation (MPC) , Distributed Learning, and Federated Learning (FL) ). These techniques have a particular focus on data encryption or secure local computation. They transfer the intermediate information to the third party to compute the final result. Gradient exchanging is commonly considered to be a secure way of training a robust model collaboratively in Deep Learning (DL) . However, recent researches have demonstrated that sensitive information can be recovered from the shared gradient. Generative Adversarial Network (GAN) , in particular, has shown to be effective in recovering such information. However, GAN based techniques require additional information, such as class labels that are generally unavailable for privacy-preserved learning. In this article, we show that, in the FL system, image-based privacy data can be easily recovered in full from the shared gradient only via our proposed Generative Regression Neural Network (GRNN) . We formulate the attack to be a regression problem and optimize two branches of the generative model by minimizing the distance between gradients. We evaluate our method on several image classification tasks. The results illustrate that our proposed GRNN outperforms state-of-the-art methods with better stability, stronger robustness, and higher accuracy. It also has no convergence requirement to the global FL model. Moreover, we demonstrate information leakage using face re-identification. Some defense strategies are also discussed in this work.

BACKGROUND: Gene prioritization (gene ranking) aims to obtain the centrality of genes, which is critical for cancer diagnosis and therapy since keys genes correspond to the biomarkers or targets of drugs. Great efforts have been devoted to the gene ranking problem by exploring the similarity between candidate and known disease-causing genes. However, when the number of disease-causing genes is limited, they are not applicable largely due to the low accuracy. Actually, the number of disease-causing genes for cancers, particularly for these rare cancers, are really limited. Therefore, there is a critical needed to design effective and efficient algorithms for gene ranking with limited prior disease-causing genes. RESULTS: In this study, we propose a transfer learning based algorithm for gene prioritization (called TLGP) in the cancer (target domain) without disease-causing genes by transferring knowledge from other cancers (source domain). The underlying assumption is that knowledge shared by similar cancers improves the accuracy of gene prioritization. Specifically, TLGP first quantifies the similarity between the target and source domain by calculating the affinity matrix for genes. Then, TLGP automatically learns a fusion network for the target cancer by fusing affinity matrix, pathogenic genes and genomic data of source cancers. Finally, genes in the target cancer are prioritized. The experimental results indicate that the learnt fusion network is more reliable than gene co-expression network, implying that transferring knowledge from other cancers improves the accuracy of network construction. Moreover, TLGP outperforms state-of-the-art approaches in terms of accuracy, improving at least 5%. CONCLUSION: The proposed model and method provide an effective and efficient strategy for gene ranking by integrating genomic data from various cancers.

Segmenting complex 3D geometry is a challenging task due to rich structural details and complex appearance variations of target object. Shape representation and foreground-background delineation are two of the core components of segmentation. Explicit shape models, such as mesh based representations, suffer from poor handling of topological changes. On the other hand, implicit shape models, such as level-set based representations, have limited capacity for interactive manipulation. Fully automatic segmentation for separating foreground objects from background generally utilizes non-interoperable machine learning methods, which heavily rely on the off-line training dataset and are limited to the discrimination power of the chosen model. To address these issues, we propose a novel semi-implicit representation method, namely Non-Uniform Implicit B-spline Surface (NU-IBS), which adaptively distributes parametrically blended patches according to geometrical complexity. Then, a two-stage cascade classifier is introduced to carry out efficient foreground and background delineation, where a simplistic Naïve-Bayesian model is trained for fast background elimination, followed by a stronger pseudo-3D Convolutional Neural Network (CNN) multi-scale classifier to precisely identify the foreground objects. A localized interactive and adaptive segmentation scheme is incorporated to boost the delineation accuracy by utilizing the information iteratively gained from user intervention. The segmentation result is obtained via deforming an NU-IBS according to the probabilistic interpretation of delineated regions, which also imposes a homogeneity constrain for individual segments. The proposed method is evaluated on a 3D cardiovascular Computed Tomography Angiography (CTA) image dataset and Brain Tumor Image Segmentation Benchmark 2015 (BraTS2015) 3D Magnetic Resonance Imaging (MRI) dataset.

Data compression forms a central role in handling the bottleneck of data storage, transmission and processing. Lossless compression requires reducing the file size whilst maintaining bit-perfect decompression, which is the main target in medical applications. This paper presents a novel lossless compression method for 16-bit medical imaging volumes. The aim is to train a neural network (NN) as a 3D data predictor, which minimizes the differences with the original data values and to compress those residuals using arithmetic coding. We evaluate the compression performance of our proposed models to state-of-the-art lossless compression methods, which shows that our approach accomplishes a higher compression ratio in comparison to JPEG-LS, JPEG2000, JP3D, and HEVC and generalizes well.

As scanners produce higher-resolution and more densely sampled images, this raises the challenge of data storage, transmission and communication within healthcare systems. Since the quality of medical images plays a crucial role in diagnosis accuracy, medical imaging compression techniques are desired to reduce scan bitrate while guaranteeing lossless reconstruction. This paper presents a lossless compression method that integrates a Recurrent Neural Network (RNN) as a 3D sequence prediction model. The aim is to learn the long dependencies of the voxel's neighbourhood in 3D using Long Short-Term Memory (LSTM) network then compress the residual error using arithmetic coding. Experiential results reveal that our method obtains a higher compression ratio achieving 15% saving compared to the state-of-the-art lossless compression standards, including JPEG-LS, JPEG2000, JP3D, HEVC, and PPMd. Our evaluation demonstrates that the proposed method generalizes well to unseen modalities CT and MRI for the lossless compression scheme. To the best of our knowledge, this is the first lossless compression method that uses LSTM neural network for 16-bit volumetric medical image compression.

In this paper, we propose a clustering approach embedded in a deep convolutional auto-encoder (DCAE). In contrast to conventional clustering approaches, our method simultaneously learns feature representations and cluster assignments through DCAEs. DCAEs have been effective in image processing as it fully utilizes the properties of convolutional neural networks. Our method consists of clustering and reconstruction objective functions. All data points are assigned to their new corresponding cluster centers during the optimization, after that, clustering centers are iteratively updated to obtain a stable performance of clustering. The experimental results on the MNIST dataset show that the proposed method substantially outperforms deep clustering models in term of clustering quality.

Very recent advances in machine learning have expanded deep learning methods to spatially-irregular data domains. Deep learning on graphs in particular has received greater study, providing benefits in numerous fields. In this paper we present a graph-based convolutional autoencoder and assess the contribution of four components towards encoding quality. A graph-based convolution-operator is used to learn localised filtering operations for graph-wise encoding. An evaluation of the proposed method is provided on a topologically-irregular version of MNIST that violates the assumption made by conventional convolutional autoencoder methods of the structure of its input-data.

The prediction of financial time series data is a challenging task due to the unpredictable behaviours of investors that are influenced by a multitude of factors. In this paper, we present a novel deep Long Short-Term Memory (LSTM) based time-series data modelling for use in stock market index prediction. A dataset comprised of six market indices from around the world were chosen to demonstrate the robustness in varying market conditions with an aim to forecast the next day closing price. With experimental results showing an average annual profitability performance of up to 200%, our method demonstrates its feasibility and significant results in time-series modelling and prediction of financial markets.

Face detection in the wild is a challenging vision problem due to large variations and unpredictable ambiguities commonly existed in real world images. Whilst introducing powerful but complex models is often computationally inefficient, using hand-crafted features is hence problematic. In this paper, we propose a nested CNN-cascade learning algorithm that adopts shallow neural network architectures that allow efficient and progressive elimination of negative hypothesis from easy to hard via self-learning discriminative representations from coarse to fine scales. The face detection problem is considered as solving three sub-problems: eliminating easy background with a simple but fast model, then localising the face region with a soft-cascade, followed by precise detection and localisation by verifying retained regions with a deeper and stronger model. The face detector is trained on the AFLW dataset following the standard evaluation procedure, and the method is tested on four other public datasets, i.e. FDDB, AFW, CMU-MIT and GENKI. Both quantitative and qualitative results on FDDB and AFW are reported, which show promising performances on detecting faces in unconstrained environment.

Face detection in the wild is a challenging vision problem due to large variations and unpredictable ambiguities commonly existed in real world images. Whilst using hand-crafted features is generally problematic, introducing powerful but complex models is often computationally inefficient. Feature aggregation and multi-resolution are two efficient strategies for traditional visual recognition methods. In this paper, we show that such strategies can be integrated into Convolutional Neural Network (CNN) architecture via average pooling and channel-wise feature concatenation. Shallow networks with feature aggregation at multi-resolution enables the traditional cascade framework to tackle the challenging detection problems efficiently. The proposed method is tested on a public benchmark with across dataset evaluation. Both quantitative and qualitative results show promising performance improvements on detecting faces in unconstrained environment.

Coronary artery segmentation plays a vital important role in coronary disease diagnosis and treatment. In this paper, we present a machine learning based interactive coronary artery segmentation method for 3D computed tomography angiography images. We first apply vessel diffusion to reduce noise interference and enhance the tubular structures in the images. A few user strokes are required to specify region of interest and background. Various image features for detecting the coronary arteries are then extracted in a multi-scale fashion, and are fed into a random forests classifier, which assigns each voxel with probability values of being coronary artery and background. The final segmentation is carried through an MRF based optimization using primal dual algorithm. A connectivity component analysis is carried out as post processing to remove isolated, small regions to produce the segmented coronary arterial vessels. The proposed method requires limited user interference and achieves robust segmentation results.

Facial fiducial point localization is a crucial step for most facial analysis applications, e.g., face recognition, expression recognition and facial aging simulation. Although state-of-art methods have the ability to provide good salient point location on frontal faces, finding a global solution under large variations caused by off-plane rotations and exaggerated expression changes is still a challenge. In this paper, we present a system with a two-level shape model to facilitate accurate facial fiducial point localization. In the first level, two local component models interact with each other in order to offer novel shape constraints. At the same time, the clamped local shape model provides constrained non-linear shape initialization for better convergence performance of the shape model as a whole. The experimental results confirm that the proposed method is capable of dealing with the face alignment under large shape variations.