Search

Domain shift refers to change of distributional characteristics between the training (source) and the testing (target) datasets of a learning task, leading to performance drop. For tasks involving medical images, domain shift may be caused because of several factors such as change in underlying imaging modalities, measuring devices and staining mechanisms. Recent approaches address this issue via generative models based on the principles of adversarial learning albeit they suffer from issues such as difficulty in training and lack of diversity. Motivated by the aforementioned observations, we adapt an alternative class of deep generative models called the Energy Based Models (EBMs) for the task of unpaired image-to-image translation of medical images. Specifically, we propose a novel method called the Boundary Preserving Twin EBMs (BPT-EBM) which employs a pair of EBMs in the latent space of an Auto-Encoder trained on the source data. While one of the EBMs translates the source to the target domain the other does vice-versa along with a novel boundary preserving loss, ensuring translation symmetry and coupling between the domains. We theoretically analyze the proposed method and show that our design leads to better translation between the domains with reduced langevin mixing steps. We demonstrate the efficacy of our method through detailed quantitative and qualitative experiments on image segmentation tasks on three different datasets vis-a-vis state-of-the-art methods.

Autism spectrum is a brain development conditionthat impairs an individual’s capacity tocommunicate socially and manifests itself throughstrict routines and obsessive-compulsive behavior.Applied behavior analysis (ABA) is thegold-standard treatment for autism spectrumdisorder (ASD). However, as the number of ASDcases increases, there is a substantial shortage oflicensed ABA practitioners, limiting the timelyformulation, revision, and implementation oftreatment plans and goals. Additionally, thesubjectivity of the clinician and a lack ofdata-driven decision-making affect treatmentquality. We address these obstacles by applying twomachine learning algorithms to recommend andpersonalize ABA treatment goals for 29 studyparticipants with ASD. The patient similarity andcollaborative filtering methods predicted ABAtreatment with an average accuracy of 81-84percent, with a normalized discounted cumulativegain of 79-81 percent (NDCG) compared toclinician-prepared ABA treatmentrecommendations. Additionally, we assess the twomodels’ treatment efficacy (TE) by measuring thepercentage of recommended treatment goalsmastered by the study participants. The proposedtreatment recommendation and personalizationstrategy are generalizable to other interventionmethods in addition to ABA and for other braindisorders. This study has been registered as aclinical trial on November 5, 2020 with trailregistration number CTRI/2020/11/028933

Voice Activity Detection (VAD) refers to the task of identification of regions of human speech in digital signals such as audio and video. While VAD is a necessary first step in many speech processing systems, it poses challenges when there are high levels of ambient noise during the audio recording. To improve the performance of VAD in such conditions, several methods utilizing the visual information extracted from the region surrounding the mouth/lip region of the speakers' video recording have been proposed. Even though these provide advantages over audio-only methods, they depend on faithful extraction of lip/mouth regions. Motivated by these, a new paradigm for VAD based on the fact that respiration forms the primary source of energy for speech production is proposed. Specifically, an audio-independent VAD technique using the respiration pattern extracted from the speakers' video is developed. The Respiration Pattern is first extracted from the video focusing on the abdominal-thoracic region of a speaker using an optical flow based method. Subsequently, voice activity is detected from the respiration pattern signal using neural sequence-to-sequence prediction models. The efficacy of the proposed method is demonstrated through experiments on a challenging dataset recorded in real acoustic environments and compared with four previous methods based on audio and visual cues., Accepted in IEEE Sensor Letters

Diagnostic and intervention methodologies for skill assessment of autism typically requires a clinician repetitively initiating several stimuli and recording the child's response. In this paper, we propose to automate the response measurement through video recording of the scene following the use of Deep Neural models for human action recognition from videos. However, supervised learning of neural networks demand large amounts of annotated data that are hard to come by. This issue is addressed by leveraging the `similarities' between the action categories in publicly available large-scale video action (source) datasets and the dataset of interest. A technique called guided weak supervision is proposed, where every class in the target data is matched to a class in the source data using the principle of posterior likelihood maximization. Subsequently, classifier on the target data is re-trained by augmenting samples from the matched source classes, along with a new loss encouraging inter-class separability. The proposed method is evaluated on two skill assessment autism datasets, SSBD and a real world Autism dataset comprising 37 children of different ages and ethnicity who are diagnosed with autism. Our proposed method is found to improve the performance of the state-of-the-art multi-class human action recognition models in-spite of supervision with scarce data., AAAI 2020

Airwriting Recognition is the problem of identifying letters written in free space with finger movement. It is essentially a specialized case of gesture recognition, wherein the vocabulary of gestures corresponds to letters as in a particular language. With the wide adoption of smart wearables in the general population, airwriting recognition using motion sensors from a smart-band can be used as a medium of user input for applications in Human-Computer Interaction. There has been limited work in the recognition of in-air trajectories using motion sensors, and the performance of the techniques in the case when the device used to record signals is changed has not been explored hitherto. Motivated by these, a new paradigm for device and user-independent airwriting recognition based on supervised contrastive learning is proposed. A two stage classification strategy is employed, the first of which involves training an encoder network with supervised contrastive loss. In the subsequent stage, a classification head is trained with the encoder weights kept frozen. The efficacy of the proposed method is demonstrated through experiments on a publicly available dataset and also with a dataset recorded in our lab using a different device. Experiments have been performed in both supervised and unsupervised settings and compared against several state-of-the-art domain adaptation techniques. Data and the code for our implementation will be made available at https://github.com/ayushayt/SCLAiR.

Objective: Since its outbreak, the rapid spread of COrona VIrus Disease 2019 (COVID-19) across the globe has pushed the health care system in many countries to the verge of collapse. Therefore, it is imperative to correctly identify COVID-19 positive patients and isolate them as soon as possible to contain the spread of the disease and reduce the ongoing burden on the healthcare system. The primary COVID-19 screening test, RT-PCR although accurate and reliable, has a long turn-around time. In the recent past, several researchers have demonstrated the use of Deep Learning (DL) methods on chest radiography (such as X-ray and CT) for COVID-19 detection. However, existing CNN based DL methods fail to capture the global context due to their inherent image-specific inductive bias. Methods: Motivated by this, in this work, we propose the use of vision transformers (instead of convolutional networks) for COVID-19 screening using the X-ray and CT images. We employ a multi-stage transfer learning technique to address the issue of data scarcity. Furthermore, we show that the features learned by our transformer networks are explainable. Results: We demonstrate that our method not only quantitatively outperforms the recent benchmarks but also focuses on meaningful regions in the images for detection (as confirmed by Radiologists), aiding not only in accurate diagnosis of COVID-19 but also in localization of the infected area. The code for our implementation can be found here - https://github.com/arnabkmondal/xViTCOS. Conclusion: The proposed method will help in timely identification of COVID-19 and efficient utilization of limited resources.

Despite their tremendous success in modelling high-dimensional data manifolds, deep neural networks suffer from the threat of adversarial attacks - Existence of perceptually valid input-like samples obtained through careful perturbation that lead to degradation in the performance of the underlying model. Major concerns with existing defense mechanisms include non-generalizability across different attacks, models and large inference time. In this paper, we propose a generalized defense mechanism capitalizing on the expressive power of regularized latent space based generative models. We design an adversarial filter, devoid of access to classifier and adversaries, which makes it usable in tandem with any classifier. The basic idea is to learn a Lipschitz constrained mapping from the data manifold, incorporating adversarial perturbations, to a quantized latent space and re-map it to the true data manifold. Specifically, we simultaneously auto-encode the data manifold and its perturbations implicitly through the perturbations of the regularized and quantized generative latent space, realized using variational inference. We demonstrate the efficacy of the proposed formulation in providing resilience against multiple attack types (black and white box) and methods, while being almost real-time. Our experiments show that the proposed method surpasses the state-of-the-art techniques in several cases. The implementation code is available at - https://github.com/mayank31398/lqvae.

In the last few years, compression of deep neural networks has become an important strand of machine learning and computer vision research. Deep models require sizeable computational complexity and storage, when used for instance for Human Action Recognition (HAR) from videos, making them unsuitable to be deployed on edge devices. In this paper, we address this issue and propose a method to effectively compress Recurrent Neural Networks (RNNs) such as Gated Recurrent Units (GRUs) and Long-Short-Term-Memory Units (LSTMs) that are used for HAR. We use a Variational Information Bottleneck (VIB) theory-based pruning approach to limit the information flow through the sequential cells of RNNs to a small subset. Further, we combine our pruning method with a specific group-lasso regularization technique that significantly improves compression. The proposed techniques reduce model parameters and memory footprint from latent representations, with little or no reduction in the validation accuracy while increasing the inference speed several-fold. We perform experiments on the three widely used Action Recognition datasets, viz. UCF11, HMDB51, and UCF101, to validate our approach. It is shown that our method achieves over 70 times greater compression than the nearest competitor with comparable accuracy for the task of action recognition on UCF11., Accepted at IEEE WACV 2021

Generative adversarial networks (GANs) have shown remarkable success in the generation of data from natural data manifolds such as images. In several scenarios, it is desirable that generated data is well-clustered, especially when there is severe class imbalance. In this paper, we focus on the problem of clustering in the generated space of GANs and uncover its relationship with the characteristics of the latent space. We derive from first principles, the necessary and sufficient conditions needed to achieve faithful clustering in the GAN framework: (i) presence of a multimodal latent space with adjustable priors, (ii) existence of a latent space inversion mechanism and, (iii) imposition of the desired cluster priors on the latent space. We also identify the GAN models in the literature that partially satisfy these conditions and demonstrate the importance of all the components required, through ablative studies on multiple real-world image datasets. Additionally, we describe a procedure to construct a multimodal latent space which facilitates learning of cluster priors with sparse supervision. Codes for our implementation is available at https://github.com/NEMGAN/NEMGAN-P .

Segmentation of the pixels corresponding to human skin is an essential first step in multiple applications ranging from surveillance to heart-rate estimation from remote-photoplethysmography. However, the existing literature considers the problem only in the visible-range of the EM-spectrum which limits their utility in low or no light settings where the criticality of the application is higher. To alleviate this problem, we consider the problem of skin segmentation from the Near-infrared images. However, Deep learning based state-of-the-art segmentation techniques demands large amounts of labelled data that is unavailable for the current problem. Therefore we cast the skin segmentation problem as that of target-independent Unsupervised Domain Adaptation (UDA) where we use the data from the Red-channel of the visible-range to develop skin segmentation algorithm on NIR images. We propose a method for target-independent segmentation where the ‘nearest-clone’ of a target image in the source domain is searched and used as a proxy in the segmentation network trained only on the source domain. We prove the existence of ‘nearest-clone’ and propose a method to find it through an optimization algorithm over the latent space of a Deep generative model based on variational inference. We demonstrate the efficacy of the proposed method for NIR skin segmentation over the state-of-the-art UDA segmentation methods on the two newly created skin segmentation datasets in NIR domain despite not having access to the target NIR data. Additionally, we report state-of-the-art results for adaption from Synthia to Cityscapes which is a popular setting in Unsupervised Domain Adaptation for semantic segmentation. The code and datasets are available at https://github.com/ambekarsameer96/GLSS.

The potential for radiomics to support oncology decision-making has grown substantially in recent years, as these scanning techniques have been found to offer unique information regarding the tumor phenotype and microenvironment that is distinct from that provided by genomic or proteomic assays. Radiomic and genomic (or proteomic) data can be correlated with one another, thereby facilitating radiogenomic efforts. Radiogenomically-informed biopsies have the potential to yield better pathological outcomes and can aid in the planning of more appropriate treatment strategies for cancer patients. However, the field lacks a unified software platform wherein radiomic and genomics/proteomic data could be brought together to conduct a variety of correlational analyses and build robust artificial intelligence models that would aid the prediction of genomic/proteomic profiles of tumors from their radiological images. We have built such a comprehensive platform that could be utilized by scientists and clinicians globally to conduct radiogenomic studies for a variety of cancer types, and further validate and deploy it in clinics to aid effective monitoring, diagnosis, and treatment of cancer patients. Citation Format: Shrey S. Sukhadia, Shivashankar H. Nagaraj, Olivier Gevaert, Sivakumaran Theru Arumugam, Aayush Tyagi, Pritam Mukherjee, A.P. Prathosh. A sophisticated bioinformatics framework for integrative study of radiomics and genomics profiles of tumors [abstract]. In: Proceedings of the AACR Virtual Special Conference on Artificial Intelligence, Diagnosis, and Imaging; 2021 Jan 13-14. Philadelphia (PA): AACR; Clin Cancer Res 2021;27(5_Suppl):Abstract nr PO-036.

Glottal Closure Instants (GCIs) correspond to the temporal locations of significant excitation to the vocal tract occurring during the production of voiced speech. GCI detection from speech signals is a well-studied problem given its importance in speech processing. Most of the existing approaches for GCI detection adopt a two-stage approach (i) Transformation of speech signal into a representative signal where GCIs are localized better, (ii) extraction of GCIs using the representative signal obtained in first stage. The former stage is accomplished using signal processing techniques based on the principles of speech production and the latter with heuristic-algorithms such as dynamic-programming and peak-picking. These methods are thus task-specific and rely on the methods used for representative signal extraction. However, in this paper, we formulate the GCI detection problem from a representation learning perspective where appropriate representation is implicitly learned from the raw-speech data samples. Specifically, GCI detection is cast as a supervised multi-task learning problem solved using a deep convolutional neural network jointly optimizing a classification and regression cost. The learning capability is demonstrated with several experiments on standard datasets. The results compare well with the state-of-the-art algorithms while performing better in the case of presence of real-world non-stationary noise., Updated submission. Figures Added. Accepted in Interspeech 2019

Clinical time series, comprising of repeated clinical measurements provide valuable information of the trajectory of patients' condition. Linear dynamical systems (LDS) are used extensively in science and engineering for modeling time series data. The observation and state variables in LDS are assumed to be uniformly sampled in time with a fixed sampling rate. The observation sequence for clinical time series is often irregularly sampled and LDS do not model such data well. In this paper, we develop two LDS-based models for irregularly sampled data. The key idea is to incorporate a temporal difference variable within the state equations of LDS whose parameters are estimated using observed data. Our models are evaluated on prediction and imputation tasks using real irregularly sampled clinical time series data and are found to outperform state-of-the-art techniques.