Research Projects
Current / Ongoing Research
Ongoing thesis work by our PhD, M.Tech., B.Tech., and Institute Post Doctoral Fellow researchers.
PhD 3
Physics-Informed Neural Networks for Time Series Prediction
Neuro-Symbolic Approaches for Explaining Deep Temporal Neural Models
Multi-modal Learning for Predictive Time Series Analysis
The work focuses on Multi-modal Learning for Predictive Time Series Analysis, with primary emphasis on time-series forecasting and secondary focus on anomaly detection. It studies how numerical time-series data can be integrated with external modalities such as text and images to capture both historical patterns and contextual information. By analysing fusion and alignment strategies at input, intermediate, and output levels, the work aims to address research gaps in multi-level alignment, advanced fusion mechanisms, and variable-aware integration of external context, leading to more accurate, reliable, and context-aware predictive models.
AI-based Procurement RFx Automation for Numaligarh Refinery Ltd - PHASE 2
In collaboration with C-DAC Bangalore and Manthhan Software Pvt. Ltd.
AI-based Procurement RFx Automation for Numaligarh Refinery Ltd - PHASE 1
LOTUS: Low-cost Innovative Technology for Water Quality Monitoring and Water Resources Management for Urban and Rural Water Systems in India
Estimation of flocculant dosage for Efficient Operation of High Rate Thickener (HRT) using Artificial Intelligence-based models
Seamless Health Monitoring and Analysis of Soldiers Using Machine Learning Approach
Estimation of Petro-Physical Properties from Seismic Attributes and Well Logs Using Advanced Artificial Intelligence
Development of Comprehensive Interactive Well Monitoring Software Tool for Drilling Supervisors
A System for Litter Monitoring and Control Using off-the-shelf Sensors and Robotic Platforms
Knowledge Graph Fusion for Time-Series Transformers
Context-Integrated Transformers for Explainable Time-Series Forecasting
Temporal Symbolic Rule Extraction from Time-Series Transformers for Post-hoc Explainability and Querying
Premium Brand Discovery for Large Marketplaces with Customer-Aware Ranking Models
Context-Driven Explainability for Transformer-Based Time-Series Prediction
A Hybrid Framework for Actionable and Physically Plausible Multivariate Time-Series Counterfactuals
Attention-Guided Counterfactual Explanations for Transformer-Based Multivariate Time-Series Classification
Counterfactual Explanations for Deep Learning-based Remaining Useful Life Prediction
Explainable Event Detection and Diagnosis for Smart Environments Using Neuro-Symbolic Methods
Context-based Failure Prognosis Using Deep Learning
Interpretation of Attention-based CNN and Transformer Neural Network for ECG-rhythm classification
Analysis of Model Compression Techniques for Detecting Atrial Fibrillation on Mobile Devices
User Activity Recognition in Multi-user Smart Environments
Rig Activity Recognition in Oil Well Drilling Operation using Artificial Intelligence-based Approaches
Reservoir Property Estimation using Neural Networks
Stuck-Pipe Problem Detection in Oil-Well Drilling Operations Using Artificial Intelligence Techniques
Estimation of porosity from Seismic Data and Well Logs using Deep Learning
Behaviour Modelling From Activity Recognition
Deep learning for Time series Prediction
Modified Condensed Nearest Neighbour for Maintenance of Case-based Reasoning System
Self-Supervised Learning-based Approaches for Computer Vision Applications
Deep Learning Models for Cardiac Abnormality Detection from ECG Signals: An Interpretability Perspective
Anomaly Detection in Oil Well Drilling Operations Using Artificial Intelligence-based Approaches
Estimation of Petro-physical Properties for Reservoir Characterisation Using Advanced Machine Learning Approaches
Non-Intrusive Human Sensing: Techniques and Applications
8
Funded Projects
3
Ongoing Theses
25
Completed Theses
High
Impact Solutions