Abstract: This disclosure relates generally to multi-class multi-label classification and more particularly to contradiction avoided learning for multi-class multi-label classification. Conventional classification methods do not consider contradictory outcomes in multi-label classification tasks wherein contradictory outcomes have significant negative impact in the classification problem solution. The present disclosure provides a contradiction avoided learning multi-class multi-label classification. The disclosed method utilizes a binary contradiction matrix constructed using domain knowledge. Based on the binary contradiction matrix the training dataset is divided into two parts, one comprising contradictions and the second without contradictions. The classification model is trained using the divided datasets using a contradiction loss and a binary cross entropy loss to avoid contradictions during learning of the classification model.

Abstract: Conventional transport mode detection relies either on GPS data or uses supervised learning for transport mode detection, requiring labelled data with hand crafted features. Embodiments of the present disclosure provide a method and system for identification of transport modes of commuters via unsupervised learning implemented using a multistage learner. Unlabeled time series data received from accelerometer of commuters mobiles from a diversified population is processed using a unique journey segment detection technique to eliminate redundant data corresponding to stationary segments of commuter or user. The non-stationary journey segments are represented using domain generalizable Invariant Auto-Encoded Compact Sequence (I-AECS), which is a learned compact representation encompassing the encoded best diversity and commonality of latent feature representation across diverse users and cities.

Abstract: Synthetic data is an annotated information that computer simulations or algorithms generate as an alternative to real-world data. synthetic data is created in digital worlds rather than collected from or measured in the real world. Embodiments herein provide a method and system for generating synthetic data with domain adaptable features using a neural network. The system is configured to receive seed data from a source domain as an input data. The seed data is considered as a normal state of a machine. The normal state, which is an initial stage of the source domain, consists of a set of features with a certain range of values. Further, a neural network based model is used to generate high quality data with adaptation of the domain specific features. To obtain large amount data for training robust deep learning models to adapt domains emphasizing set of features/providing higher importance selectively.

Abstract: Embodiments of the present disclosure provide a method and system for co-operative and cascaded inference on the edge device using an integrated Deep Learning (DL) model for object detection and localization, which comprises a strong classifier trained on largely available datasets and a weak localizer trained on scarcely available datasets, and work in coordination to first detect object (fire) in every input frame using the classifier, and then trigger a localizer only for the frames that are classified as fire frames. The classifier and the localizer of the integrated DL model are jointly trained using Multitask Learning approach. Works in literature hardly address the technical challenge of embedding such integrated DL model to be deployed on edge devices. The method provides an optimal hardware software partitioning approach for components or segments of the integrated DL model which achieves a tradeoff between latency and accuracy in object classification and localization.

Abstract: Conventional gesture detection approaches demand large memory and computation power to run efficiently, thus limiting their use in power and memory constrained edge devices. Present application/disclosure provides a Spiking Neural Network based system which is a robust low power edge compatible ultrasound-based gesture detection system. The system uses a plurality of speakers and microphones that mimics a Multi Input Multi Output (MIMO) setup thus providing requisite diversity to effectively address fading. The system also makes use of distinctive Channel Impulse Response (CIR) estimated by imposing sparsity prior for robust gesture detection. A multi-layer Convolutional Neural Network (CNN) has been trained on these distinctive CIR images and the trained CNN model is converted into an equivalent Spiking Neural Network (SNN) via an ANN (Artificial Neural Network)-to-SNN conversion mechanism. The SNN is further configured to detect/classify gestures performed by user(s).

Abstract: Model-based image reconstruction (MBIR) methods using convolutional neural networks (CNNs) as priors have demonstrated superior image quality and robustness compared to conventional methods. Studies have explored MBIR combined with supervised and unsupervised denoising techniques for image reconstruction in magnetic resonance imaging (MRI) and positron emission tomography (PET). Unsupervised methods like the deep image prior (DIP) have shown promising results and are less prone to hallucinations. However, since the noisy image is used as a reference, strategies to prevent overfitting are unclear. Recently, Bayesian DIP (BDIP) networks that model uncertainty tend to prevent overfitting without requiring early stopping. However, BDIP has not been studied with data-fidelity term for image reconstruction. Present disclosure provides systems and method that implement a MBIR framework with a modified BDIP.

Abstract: This disclosure relates generally to the field of Electroencephalogram (EEG) classification, and, more particularly, to method and system for EEG motor imagery classification. Existing deep learning works employ the sensor-space for EEG graph representations wherein the channels of the EEG are considered as nodes and connection between the nodes are either predefined or are based on certain heuristics. However, these representations are ineffective and fail to accurately capture the underlying brain's functional networks. Embodiments of present disclosure provide a method of training a weighted adjacency matrix and a Graph Neural Network (GNN) to accurately represent the EEG signals. The method also trains a graph, a node, and an edge classifier to perform graph classification (i.e. motor imagery classification), node and edge classification. Thus, representations generated by the GNN can be additionally used for node and edge classification unlike state of the art methods.

Abstract: The present invention relates to the field of evaluating clinical diagnostic models. Conventional metrics does not consider context dependent clinical principles and is unable to capture critically important features that ought to be present in a diagnostic model. Thus, present disclosure provides a method and system for evaluating clinical efficacy of multi-label multi-class computational diagnostic models. Diagnosis for a given dataset of diagnostic samples is obtained from the diagnostic model which is then classified as wrong, missed, over or right diagnosis, based on which a first penalty is calculated. A second penalty is calculated for each diagnostic sample using a contradiction matrix. The first and second penalties are summed up to compute a pre-score for each diagnostic sample. Finally, the diagnostic model is evaluated using a metric that is based on sum of pre-scores, and scores from a perfect and a null multi-label multi-class computational diagnostic model.

Abstract: Portable ECG monitors available in market have the disadvantage that the ECG data they provide as input aren't directly interpretable and requires medical knowledge for the users. The disclosure herein generally relates to Electrocardiogram (ECG), and, more particularly, to a method and system for generating 2d representation of electrocardiogram (ECG) signals. The system provides a mechanism for determining variability between a plurality of segments of an ECG data measured, and uses the information on the determined variability to generate the 2D representation corresponding to the ECG signal. The system further provides means to generate a data model that can be further used for processing real-time ECG data for generating corresponding interpretations. This allows a user to obtain the interpretations as output.

Abstract: The disclosure relates generally to methods and systems for monitoring lubricant oil condition using a photoacoustic modelling. Conventional techniques in the art for checking the condition of the lubricant oil is laboratory based and thus time consuming, error prone and not efficient. The present disclosure discloses a photoacoustic simulation model which is developed utilizing a photonic model such as a Monte Carlo method-based optical simulation integrated with a finite element model such as a k-wave toolbox-based acoustic measurement. The photoacoustic simulation model of the present disclosure is used to obtain a photoacoustic signal of the lubricant oil sample and a set of statistical features are determined from the obtained photoacoustic signal. The determined set of statistical features are then used as a training data to develop a machine learning (ML) model which is used to classify a type of contamination of the test lubricating oil.

Abstract: This disclosure relates to systems and methods for shapelet decomposition based recognition using radar. State-of-the-art solutions involve use of standard machine learning classification techniques for gesture recognition which suffer with problem of dependency on collected data. The present disclosure overcome the limitations faced by the state-of-the-art solutions by obtaining a plurality of time domain signal using a radar system comprising three vertically arranged radars and one or more sensors, identifying one or more gesture windows to obtain one or more shapelets corresponding to one or gestures which are further decomposed into a plurality of sub shapelets. Further, at least one of (i) a positive or (i) a negative time delay is applied to each of the plurality of sub shapelets to obtain a plurality of composite shapelets which are further mapped with a plurality of trained shapelets to recognize gestures comprised in one or more activities performed by a subject.

Abstract: One of the biggest challenges faced by oil and gas companies is to monitor such long pipelines for leak events and generate false leak event alarms during routine pipe maintenance. A data associated with a first sensing unit is processed to obtain an instant timing information (T0) of a leak event in a conduit at a test environment. A data associated with a second sensing unit is processed to obtain a transient signal associated with the leak event at a specific band. An accelerometer data is filtered to obtain a band passed filtered accelerometer signal (Accelbpf). The Accelbpf is truncated in a time domain from the T0 to a duration Td of the leak event to obtain a temporal template signal (Acceltemplate). A leak event of a real-time conduit is dynamically detected at a physical environment based on Acceltemplate when a cross-correlation value is greater than a threshold value (?).

Abstract: This disclosure relates generally to method and system for jointly pruning and hardware acceleration of pre-trained deep learning models. The present disclosure enables pruning a plurality of DNN models layers using an optimal pruning ratio. The method processes a pruning request to transform the plurality of DNN models and the plurality of hardware accelerators into a plurality of pruned hardware accelerated DNN models based on at least one user option. The first pruning search option executes a hardware pruning search technique to perform search on each DNN model and each processor based on at least one of a performance indicator and an optimal pruning ratio. The second pruning search option executes an optimal pruning search technique, to perform search on each layer with corresponding pruning ratio.

Abstract: This disclosure relates generally to methods and systems for unobtrusive digital health assessment of high risk subjects, wherein bio-markers pertaining to a disease are identified automatically using physical activity and physiology monitoring on a continuous basis. Identification of bio-markers in the medical domain is conventionally dependent on insights derived from medical tests which are obtrusive in nature. Systems and methods of the present disclosure integrate physical characteristics, lifestyle habits and prevailing medical conditions with monitored physical activities and physiological measurements to assess health of high risk subjects. Systems and methods of the present disclosure also enable automatic generation of control class and treatment class that may be effectively used for health assessment.

Abstract: The disclosure generally relates to scene graph generation. Scene graph captures rich semantic information of an image by representing objects and their relationships as nodes and edges of a graph and has several applications including image retrieval, action recognition, visual question answering, autonomous driving, robotics. However, to leverage scene graphs, computationally efficient scene graph generation methods are required, which is very challenging to generate due presence of a quadratic number of potential edges and computationally intensive/non-scalable techniques for detecting the relationship between each object pair using the traditional approach. The disclosure proposes a combination of edge proposal neural network and the Graph neural network with spatial message passing (GNN-SMP) along with several techniques including a feature extraction technique, object detection technique, un-labelled graph generation technique and a scene graph generation technique to generate scene graphs.

Abstract: This disclosure relates generally to multi-port multi-functional meta-surface coplanar antenna system. Conventional electronic or mechanical solutions for beam steering incur high installation costs with less performance speed and bulk structures. The present disclosure provides multi-port multi-functional meta-surface coplanar antenna system for beam steering control. The disclosed antenna system enables radiator to have a performance diversity application through beam steering functionalities. The disclosed antenna system provides a minimal design complexity and minimal usage of active or passive lumped components. The disclosed system comprises Gradient Refractive Index Meta-surface (GRIM) and the antenna disposed on the same side of a substrate. Beam steering control is performed using port excitations and controlling the phase between the concerned ports externally.

Abstract: Direct usage of endosomatic EDA has multiple challenges for practical cognitive load assessment. Embodiments of the method and system disclosed provide a solution to the technical challenges in the art by directly using the bio-potential signals to implement endosomatic approach for assessment of cognitive load. The method utilizes a multichannel wearable endosomatic device capable of acquiring and combining multiple bio-potentials, which are biomarkers of cognitive load experienced by a subject performing a cognitive task. Further, extracts information for classification of the cognitive load, from the acquired bio-signals using a set of statistical and a set of spectral features. Furthermore, utilizes a feature selection approach to identify a set of optimum features to train a Machine Learning (ML) based task classifier to classify the cognitive load experienced by a subject into high load task and low load task.

Abstract: This disclosure relates generally to identification and mitigation of bias while training deep learning models. Conventional methods do not provide effective methods for bias identification, and they require pre-defined concepts and rules for bias mitigation. The embodiments of the present disclosure train an auto-encoder to produce a generalized representation of an input image by decomposing into a set of latent embedding. The set of latent embedding are used to learn the shape and color concepts of the input image. The feature specialization is done by training an auto-encoder to reconstruct the input image using the shape embedding modulated by color embedding. To identify the bias, permutation invariant neural network is trained for classification task and attribution scores corresponding to each concept embedding are computed. The method also performs de-biasing the classifier by training it with a set of counterfactual images generated by modifying the latent embedding learned by the auto-encoder.

Abstract: The present invention relates to a method and system for early detection of COVID-19. Existing methods require data from multiple sensors for training a prediction model whose output is considered as final prediction which is actually the prediction for a particular day or time instance. However, this prediction doesn't detect actual infection of COVID-19 since it requires monitoring the change in health of the user over consecutive days. Embodiments of present disclosure overcome these challenges by a prediction model for COVID-19 which requires only data from Photoplethysmography (PPG) sensor seamlessly collected from a wearable device still able to provide accurate COVID-19 prediction with application of a post processing technique on the predictions of the prediction model. Since COVID-19 symptoms have an effect on heartrate and oxygen saturation which are effectively captured by PPG sensor data, studying these dynamics during infection period gives insights to perform early detection of COVID-19.

Abstract: Optical images in remote sensing are contaminated by cloud cover and bad weather conditions and are only available during the daytime. Whereas SAR images are completely cloud free, independent of weather conditions and can be acquired both during the day and at night. However, due to the speckle effect and side looking imaging mechanism of SAR images, they are not easily interpretable by untrained people. To address this issue, the present disclosure provides a method and system for LULC guided SAR visualization, wherein a GAN is trained to translate SAR images to optical images for visualization. A given SAR image is fed into a first generator of the GAN to obtain LULC map which is then concatenated with the SAR image and fed into a second generator of the GAN to generate an optical image. The LULC map provides semantic information required for generation of more realistic optical image.