Tabular data is the most abundant data type in the world, powering systems in finance, healthcare, e‑commerce, and beyond. As tabular datasets grow and span multiple related targets, there is an increasing need to exploit shared task information for improved multitask generalization. Multitask learning (MTL) has emerged as a powerful way to improve generalization and efficiency, yet most existing work focuses narrowly on large‑scale recommendation systems, leaving its potential in broader tabular domains largely underexplored. Also, existing MTL approaches for tabular data predominantly rely on multi-layer perceptron-based backbones, which struggle to capture complex feature interactions and often fail to scale when data is abundant, a limitation that transformer architectures have overcome in other domains. Motivated by this, we introduce MultiTab-Net, the first multitask transformer architecture specifically designed for large tabular data. MultiTab-Net employs a novel multitask masked‑attention mechanism that dynamically models feature–feature dependencies while mitigating task competition. Through extensive experiments, we show that MultiTab-Net consistently achieves higher multitask gain than existing MTL architectures and single‑task transformers across diverse domains including large‑scale recommendation data, census‑like socioeconomic data, and physics datasets, spanning a wide range of task counts, task types, and feature modalities. In addition, we contribute MultiTab-Bench, a generalized multitask synthetic dataset generator that enables systematic evaluation of multitask dynamics by tuning task count, task correlations, and relative task complexity.

Mainstream unsupervised anomaly detection algorithms often excel in academic datasets, yet their real-world performance is restricted due to the controlled experimental conditions involving clean training data. Addressing the challenge of training with noise, a prevalent issue in practical anomaly detection, is frequently overlooked. In a pioneering endeavor, this study delves into the realm of label-level noise within sensory time-series anomaly detection (TSAD). This paper presents a novel and practical TSAD when the training data is contaminated with anomalies. The introduced approach, called TSAD-C, is devoid of access to abnormality labels during the training phase. TSAD-C encompasses three modules: a Decontaminator to rectify anomalies present during training and swiftly prepare the decontaminated data for subsequent modules; a Long-range Variable Dependency Modeling module to capture long-range intra- and inter-variable dependencies within the decontaminated data that is considered as a surrogate of the pure normal data; and an Anomaly Scoring module that leverages insights of the first two modules to detect all types of anomalies. Our extensive experiments conducted on four reliable, diverse, and challenging datasets conclusively demonstrate that TSAD-C surpasses existing methods, thus establishing a new state-of-the-art in the TSAD field.

Skeleton-based video anomaly detection (SVAD) is a crucial task in computer vision. Accurately identifying abnormal patterns or events enables operators to promptly de-tect suspicious activities, thereby enhancing safety. Achieving this demands a comprehensive understanding of human motions, both at body and region levels, while also accounting for the wide variations of performing a single action. However, existing studies fail to simultaneously address these crucial properties. This paper introduces a novel, practical, and lightweight framework, namely Graph-Jigsaw Conditioned Diffusion Model for Skeleton-based Video Anomaly Detection (GiCiSAD) to overcome the challenges associated with SVAD. GiCiSAD consists of three novel modules: the Graph Attention-based Forecasting module to capture the spatio-temporal dependencies inherent in the data, the Graph-level Jigsaw Puzzle Maker module to distinguish subtle region-level discrepancies between normal and abnormal motions, and the Graph-based Conditional Diffusion model to generate a wide spectrum of human motions. Extensive experiments on four widely used skeleton-based video datasets show that GiCiSAD outperforms existing methods with significantly fewer training parameters, establishing it as the new state-of-the-art.

Multitask learning (MTL) has become prominent for its ability to predict multiple tasks jointly, achieving better per-task performance with fewer parameters than single-task learning. Recently, decoder-focused architectures have sig-nificantly improved multitask performance by refining task predictions using features from related tasks. However, most refinement methods struggle to efficiently capture both local and long-range dependencies between task-specific representations and cross-task patterns. In this paper, we introduce the Cross-Task Affinity Learning (CTAL) module, a lightweight framework that enhances task refinement in multitask networks. CTAL effectively captures local and long-range cross-task interactions by optimizing task affin-ity matrices for parameter-efficient grouped convolutions without concern for information loss. Our results demon-strate state-of-the-art MTL performance for both CNN and transformer backbones, using significantly fewer parame-ters than single-task learning.

With the recent advances in technology, a wide range of systems continue to collect a large amount of data over time and thus generate time series. Time-Series Anomaly Detection (TSAD) is an important task in various time-series applications such as e-commerce, cybersecurity, vehicle maintenance, and healthcare monitoring. However, this task is very challenging as it requires considering both the intra-variable dependency (relationships within a variable over time) and the inter-variable dependency (relationships between multiple variables) existing in time-series data. Recent graph-based approaches have made impressive progress in tackling the challenges of this field. In this survey, we conduct a comprehensive and up-to-date review of TSAD using graphs, referred to as G-TSAD. First, we explore the significant potential of graph representation for time-series data and and its contributions to facilitating anomaly detection. Then, we review state-of-the-art graph anomaly detection techniques, mostly leveraging deep learning architectures, in the context of time series. For each method, we discuss its strengths, limitations, and the specific applications where it excels. Finally, we address both the technical and application challenges currently facing the field, and suggest potential future directions for advancing research and improving practical outcomes.

Human activity recognition (HAR) is a popular research field in computer vision that has already been widely studied. However, it is still an active research field since it plays an important role in many current and emerging real-world intelligent systems, like visual surveillance and human-computer interaction. Deep reinforcement learning (DRL) has recently been used to address the activity recognition problem with various purposes, such as finding attention in video data or obtaining the best network structure. DRL-based HAR has only been around for a short time, and it is a challenging, novel field of study. Therefore, to facilitate further research in this area, we have constructed a comprehensive survey on activity recognition methods that incorporate DRL. Throughout the article, we classify these methods according to their shared objectives and delve into how they are ingeniously framed within the DRL framework. As we navigate through the survey, we conclude by shedding light on the prominent challenges and lingering questions that await the attention of future researchers, paving the way for further advancements and breakthroughs in this exciting domain.

Numerous methods for time-series anomaly detection (TSAD) have emerged in recent years, most of which are unsupervised and assume that only normal samples are available during the training phase, due to the challenge of obtaining abnormal data in real-world scenarios. Still, limited samples of abnormal data are often available, albeit they are far from representative of all possible anomalies. Supervised methods can be utilized to classify normal and seen anomalies, but they tend to overfit to the seen anomalies present during training, hence, they fail to generalize to unseen anomalies. We propose the first algorithm to address the open-set TSAD problem, called Multivariate Open-Set time-series Anomaly Detector (MOSAD), that leverages only a few shots of labeled anomalies during the training phase in order to achieve superior anomaly detection performance compared to both supervised and unsupervised TSAD algorithms. MOSAD is a novel multi-head TSAD framework with a shared representation space and specialized heads, including the Generative head, the Discriminative head, and the Anomaly-Aware Contrastive head. The latter produces a superior representation space for anomaly detection compared to conventional supervised contrastive learning. Extensive experiments on three real-world datasets establish MOSAD as a new state-of-the-art in the TSAD field.

One-Class anomaly detection aims to detect anomalies from normal samples using a model trained on normal data. With recent advancements in deep learning, researchers have designed efficient one-class anomaly detection methods. Existing works commonly use neural networks to map the data into a more informative representation and then apply an anomaly detection algorithm. In this paper, we propose a method, DASVDD, that jointly learns the parameters of an autoencoder while minimizing the volume of an enclosing hypersphere on its latent representation. We propose a novel anomaly score that combines the autoencoder's reconstruction error and the distance from the center of the enclosing hypersphere in the latent representation. Minimizing this anomaly score aids us in learning the underlying distribution of the normal class during training. Including the reconstruction error in the anomaly score ensures that DASVDD does not suffer from the hypersphere collapse issue since the DASVDD model does not converge to the trivial solution of mapping all inputs to a constant point in the latent representation. Experimental evaluations on several benchmark datasets show that the proposed method outperforms the commonly used state-of-the-art anomaly detection algorithms while maintaining robust performance across different anomaly classes.

Deep clustering incorporates embedding into clustering in order to find a lower-dimensional space suitable for clustering tasks. Conventional deep clustering methods aim to obtain a single global embedding subspace (aka latent space) for all the data clusters. In contrast, in this article, we propose a deep multirepresentation learning (DML) framework for data clustering whereby each difficult-to-cluster data group is associated with its own distinct optimized latent space and all the easy-to-cluster data groups are associated with a general common latent space. Autoencoders (AEs) are employed for generating cluster-specific and general latent spaces. To specialize each AE in its associated data cluster(s), we propose a novel and effective loss function which consists of weighted reconstruction and clustering losses of the data points, where higher weights are assigned to the samples more probable to belong to the corresponding cluster(s). Experimental results on benchmark datasets demonstrate that the proposed DML framework and loss function outperform state-of-the-art clustering approaches. In addition, the results show that the DML method significantly outperforms the SOTA on imbalanced datasets as a result of assigning an individual latent space to the difficult clusters.

Electroencephalogram (EEG) signals are effective tools towards seizure analysis where one of the most important challenges is accurate detection of seizure events and brain regions in which seizure happens or initiates. However, all existing machine learning-based algorithms for seizure analysis require access to the labeled seizure data while acquiring labeled data is very labor intensive, expensive, as well as clinicians dependent given the subjective nature of the visual qualitative interpretation of EEG signals. In this paper, we propose to detect seizure channels and clips in a self-supervised manner where no access to the seizure data is needed. The proposed method considers local structural and contextual information embedded in EEG graphs by employing positive and negative sub-graphs. We train our method through minimizing contrastive and generative losses. The employ of local EEG sub-graphs makes the algorithm an appropriate choice when accessing to the all EEG channels is impossible due to complications such as skull fractures. We conduct an extensive set of experiments on the largest seizure dataset and demonstrate that our proposed framework outperforms the state-of-the-art methods in the EEG-based seizure study. The proposed method is the only study that requires no access to the seizure data in its training phase, yet establishes a new state-of-the-art to the field, and outperforms all related supervised methods.

Deep learning-based algorithms have been very successful in skeleton-based human activity recognition. Skeleton data contains 2-D or 3-D coordinates of human body joints. The main focus of most of the existing skeleton-based activity recognition methods is on designing new deep architectures to learn discriminative features, where all body joints are considered equally important in recognition. However, the importance of joints varies as an activity proceeds within a video and across different activities. In this work, we hypothesize that selecting relevant joints, prior to recognition, can enhance performance of the existing deep learning-based recognition models. We propose a spatial hard attention finding method that aims to remove the uninformative and/or misleading joints at each frame. We formulate the joint selection problem as a Markov decision process and employ deep reinforcement learning to train the proposed spatial-attention-aware agent. No extra labels are needed for the agent’s training. The agent takes a sequence of features extracted from skeleton video as input and outputs a sequence of probabilities for joints. The proposed method can be considered as a general framework that can be integrated with the existing skeleton-based activity recognition methods for performance improvement purposes. We obtain very competitive activity recognition results on three commonly used human activity recognition datasets.

We propose an acoustic anomaly detection algorithm based on the framework of contrastive learning. Contrastive learning is a recently proposed self-supervised approach that has shown promising results in image classification and speech recognition. However, its application in anomaly detection is underexplored. Earlier studies have demonstrated that it can achieve state-of-the-art performance in image anomaly detection, but its capability in anomalous sound detection is yet to be investigated. For the first time, we propose a contrastive learning-based framework that is suitable for acoustic anomaly detection. Since most existing contrastive learning approaches are targeted toward images, the effect of other data transformations on the performance of the algorithm is unknown. Our framework learns a representation from unlabeled data by applying audio-specific data augmentations. We show that in the resulting latent space, normal and abnormal points are distinguishable. Experiments conducted on the MIMII dataset confirm that our approach can outperform competing methods in detecting anomalies.

Multitask learning (MTL) has recently gained a lot of popularity as a learning paradigm that can lead to improved per-task performance while also using fewer per-task model parameters compared to single task learning. One of the biggest challenges regarding MTL networks involves how to share features across tasks. To address this challenge, we propose the Attentive Task Interaction Network (ATI-Net). ATI-Net employs knowledge distillation of the latent features for each task, then combines the feature maps to provide improved contextualized information to the decoder. This novel approach to introducing knowledge distillation into an attention based multitask network outperforms state of the art MTL baselines such as the standalone MTAN and PAD-Net, with roughly the same number of model parameters.

Deep clustering algorithms utilize a deep neural network to map data points in a lower-dimensional space which is more suitable for clustering task. Recent algorithms employ autoencoder to jointly learn a lower-dimensional space (aka latent space) and perform data clustering through minimizing a clustering loss. These algorithms suffer from the fact that the true cluster assignments are unknown because of the unsupervised nature of the task. Thus, they adopt a self-training strategy and estimate the true cluster labels using the algorithm parameters; while the true parameters’ value is unknown at the problem outset. To address this difficulty, we propose a deep clustering technique, called IDECF, whereby the true cluster assignments are estimated using an individual deep fully connected network (FCM-Net) which takes its input from the latent space of an autoencoder. The proposed IDECF is trained in an end-to-end manner by minimizing a linear combination of reconstruction loss and clustering loss. Experimental results on benchmark datasets demonstrate the viability and effectiveness of the proposed algorithm.

Deep clustering combines embedding and clustering together to obtain an optimal low dimensional embedding sub- space (aka latent subspace) for clustering, which can be more effective compared to conventional clustering approaches such as k-means. Typical deep clustering methods employ autoencoder (AE) and obtain their optimal latent space through minimizing data reconstruction loss which has no substantial connection with the clustering performance. In contrast, in this paper we propose a novel AE-based clustering scheme Deep Successive Subspace Learning (DSSL) which simultaneously minimizes weighted reconstruction and clustering losses of data points, where weights are defined based on similarity between latent representation of data points and cluster centers. DSSL obtains its optimal latent space through K (i.e. number of clusters) successive training runs where each run corresponds to an individual cluster. At each run, DSSL focuses on reconstruction and clustering of those data points that are more likely to belong to the corresponding cluster; hence, implicitly training those network parameters that have more influence on that cluster. Experimental results on benchmark datasets demonstrate that the proposed DSSL method can significantly outperform state-of-the-art clustering approaches.

Skeleton-based human activity recognition has attracted lots of attention due to its wide range of applications. Skeleton data includes two or three dimensional coordinates of body joints. Not all of the joints are effective in recognizing an activity. In this paper, we propose a novel framework for identifying relevant joints per activity and using them for the purpose of activity recognition. We propose to formulate the joint selection problem as a Markov Decision Process (MDP) where we employ deep reinforcement learning to find the most informative joints per frame. The proposed joint selection method is a general framework that can be employed to improve the existing human activity classification methods. Experimental results on two benchmark activity recognition data sets using three different classifiers demonstrate the effectiveness of the proposed joint selection method.

In many real-world scenarios, data from multiple modalities (sources) are collected during a development phase. Such data are referred to as multiview data. While additional information from multiple views often improves the performance, collecting data from such additional views during the testing phase may not be desired due to the high costs associated with measuring such views or, unavailability of such additional views. Therefore, in many applications, despite having a multiview training data set, it is desired to do performance testing using data from only one view. In this paper, we present a multiview feature selection method that leverages the knowledge of all views and use it to guide the feature selection process in an individual view. We realize this via a multiview feature weighting scheme such that the local margins of samples in each view are maximized and similarities of samples to some reference points in different views are preserved. Also, the proposed formulation can be used for cross-view matching when the view-specific feature weights are pre-computed on an auxiliary data set. Promising results have been achieved on nine real-world data sets as well as three biometric recognition applications. On average, the proposed feature selection method has improved the classification error rate by 31 percent of the error rate of the state-of-the-art.

Mismatch negativity (MMN) is a component of the event-related potential (ERP) that is elicited through an odd-ball paradigm. The existence of the MMN in a coma patient has a good correlation with coma emergence; however, this component can be difficult to detect. Previously, MMN detection was based on visual inspection of the averaged ERPs by a skilled clinician, a process that is expensive and not always feasible in practice. In this paper, we propose a practical machine learning (ML) based approach for detection of MMN component, thus, improving the accuracy of prediction of emergence from coma. Furthermore, the method can operate on an automatic and continuous basis thus alleviating the need for clinician involvement. The proposed method is capable of the MMN detection over intervals as short as two minutes. This finer time resolution enables identification of waxing and waning cycles of a conscious state. An auditory odd-ball paradigm was applied to 22 healthy subjects and 2 coma patients. A coma patient is tested by measuring the similarity of the patient's ERP responses with the aggregate healthy responses. Because the training process for measuring similarity requires only healthy subjects, the complexity and practicality of training procedure of the proposed method are greatly improved relative to training on coma patients directly. Since there are only two coma patients involved with this study, the results are reported on a very preliminary basis. Preliminary results indicate we can detect the MMN component with an accuracy of 92.7% on healthy subjects. The method successfully predicted emergence in both coma patients when conventional methods failed. The proposed method for collecting training data using exclusively healthy subjects is a novel approach that may prove useful in future, unrelated studies where ML methods are used.

Electrocardiogram (ECG) and transient evoked otoacoustic emission (TEOAE) are among the physiological signals that have attracted significant interest in biometric community due to their inherent robustness to replay and falsification attacks. However, they are time-dependent signals and this makes them hard to deal with in across-session human recognition scenario where only one session is available for enrollment. This paper presents a novel feature selection method to address this issue. It is based on an auxiliary dataset with multiple sessions where it selects a subset of features that are more persistent across different sessions. It uses local information in terms of sample margins while enforcing an across-session measure. This makes it a perfect fit for aforementioned biometric recognition problem. Comprehensive experiments on ECG and TEOAE variability due to time lapse and body posture are done. Performance of the proposed method is compared against seven state-of-the-art feature selection algorithms as well as another six approaches in the area of ECG and TEOAE biometric recognition. Experimental results demonstrate that the proposed method performs noticeably better than other algorithms.

Fingerprints have been extensively used for biometric recognition around the world. However, fingerprints are not secrets, and an adversary can synthesis a fake finger to spoof the biometric system. The mainstream of the current fingerprint spoof detection methods are basically binary classifier trained on some real and fake samples. While they perform well on detecting fake samples created by using the same methods used for training, their performance degrades when encountering fake samples created by a novel spoofing method. In this paper, we approach the problem from a different perspective by incorporating electrocardiogram (ECG). Compared with the conventional biometrics, stealing someone's ECG is far more difficult if not impossible. Considering that ECG is a vital signal and motivated by its inherent liveness, we propose to combine it with a fingerprint liveness detection algorithm. The combination is natural as both ECG and fingerprints can be captured from fingertips. In the proposed framework, the ECG and fingerprint are combined not only for authentication purpose but also for liveness detection. We also examine automatic template updating using ECG and fingerprint. In addition, we propose a stopping criterion that reduces the average waiting time for signal acquisition. We have performed extensive experiments on the LivDet2015 database which is presently the latest available liveness detection database and compare the proposed method with six liveness detection methods as well as 12 participants of LivDet2015 competition. The proposed system has achieved a liveness detection equal error rate (EER) of 4.2% incorporating only 5 s of ECG. By extending the recording time to 30 s, liveness detection EER reduces to 2.6% which is about 4 times better than the best of six comparison methods. This is also about 2 times better than the best results achieved by the participants of the LivDet2015 competition.

Conventional feature selection algorithms assign a single common feature set to all regions of the sample space. In contrast, this paper proposes a novel algorithm for localized feature selection for which each region of the sample space is characterized by its individual distinct feature subset that may vary in size and membership. This approach can therefore select an optimal feature subset that adapts to local variations of the sample space, and hence offer the potential for improved performance. Feature subsets are computed by choosing an optimal coordinate space so that, within a localized region, within-class distances and between-class distances are, respectively, minimized and maximized. Distances are measured using a logistic function metric within the corresponding region. This enables the optimization process to focus on a localized region within the sample space. A local classification approach is utilized for measuring the similarity of a new input data point to each class. The proposed logistic localized feature selection (lLFS) algorithm is invariant to the underlying probability distribution of the data; hence, it is appropriate when the data are distributed on a nonlinear or disjoint manifold. lLFS is efficiently formulated as a joint convex/increasing quasi-convex optimization problem with a unique global optimum point. The method is most applicable when the number of available training samples is small. The performance of the proposed localized method is successfully demonstrated on a large variety of data sets. We demonstrate that the number of features selected by the lLFS method saturates at the number of available discriminative features. In addition, we have shown that the Vapnik-Chervonenkis dimension of the localized classifier is finite. Both these factors suggest that the lLFS method is insensitive to the overfitting issue, relative to other methods.

Typical feature selection methods choose an optimal global feature subset that is applied over all regions of the sample space. In contrast, in this paper we propose a novel localized feature selection (LFS) approach whereby each region of the sample space is associated with its own distinct optimized feature set, which may vary both in membership and size across the sample space. This allows the feature set to optimally adapt to local variations in the sample space. An associated method for measuring the similarities of a query datum to each of the respective classes is also proposed. The proposed method makes no assumptions about the underlying structure of the samples; hence the method is insensitive to the distribution of the data over the sample space. The method is efficiently formulated as a linear programming optimization problem. Furthermore, we demonstrate the method is robust against the over-fitting problem. Experimental results on eleven synthetic and real-world data sets demonstrate the viability of the formulation and the effectiveness of the proposed algorithm. In addition we show several examples where localized feature selection produces better results than a global feature selection method.

Methods for pressure sore monitoring remain both a clinical and research challenge. Improved methodologies could assist physicians in developing prompt and effective pressure sore interventions. In this paper a technique is introduced for the assessment of pressure sores in guinea pigs, using captured color images. Sores were artificially induced, utilizing a system particularly developed for this purpose. Digital images were obtained from the suspicious region in days 3 and 7 post-pressure sore generation. Different segments of the color images were divided and labeled into three classes, based on their severity status. For quantitative analysis, a color based texture model, which is invariant against monotonic changes in illumination, is proposed. The texture model has been developed based on the local binary pattern operator. Tissue segments were classified, using the texture model and its features as inputs to a combination of neural networks. Our method is capable of discriminating tissue segments in different stages of pressure sore generation, and therefore can be a feasible tool for the early assessment of pressure sores.