Publications & Patents
Publications & Patents
Authors: Chathurangi Shyalika, Harleen Kaur Bagga, Ahan Bhatt, Renjith Prasad, Alaa Al Ghazo, Amit Sheth (2025)
Publication date: December 2024
Conference : AAAI -2025
Abstract: Time series foundational models (TSFM) have gained prominence in time series forecasting, promising state-of-the-art performance across various applications. However, their application in anomaly detection and prediction remains underexplored, with growing concerns regarding their black-box nature, lack of interpretability and applicability. This paper critically evaluates the efficacy of TSFM in anomaly detection and prediction tasks. We systematically analyze TSFM across multiple datasets, including those characterized by the absence of discernible patterns, trends and seasonality. Our analysis shows that while TSFMs can be extended for anomaly detection and prediction, traditional statistical and deep learning models often match or outperform TSFM in these tasks. Additionally, TSFMs require high computational resources but fail to capture sequential dependencies effectively or improve performance in few-shot or zero-shot scenarios. The preprocessed datasets, codes to reproduce the results and supplementary materials are available at this https URL.
MultiAgent Neurosymbolic system for Smart Manufacturing
SmartPilot: Agent-Based CoPilot for Intelligent Manufacturing
Authors: Author1, Chathurangi Shyalika, Author2, Renjith Prasad, Author3, Alaa Al Ghazo, Author4, Harleen Kaur Bagga(2025)
Publication date: March 2025
Conference- IEEE Conference on Artificial Intelligence 2025
Abstract: In the dynamic landscape of Industry 4.0, achieving efficiency, precision, and adaptability is essential to optimize manufacturing operations. Industries suffer due to supply chain distruptions caused by anomalies, which are being detected by current AI models but leaving the domain experts uncertain without deeper insights into these anomalies. Additionally, operational inefficiencies persist due to inaccurate production forecasts and the limited effectiveness of traditional AI models for processing complex sensor data. Despite these advancements, existing systems lack the seamless integration of these capabilities needed to create a truly unified solution for enhancing production and decision-making. We propose SmartPilot, a neurosymbolic, multiagent CoPilot designed for advanced reasoning and contextual decision-making to address these challenges. SmartPilot processes multimodal sensor data and is compact to deploy on edge devices. It focuses on three key tasks: anomaly prediction, production forecasting, and domain-specific question answering. By bridging the gap between AI capabilities and real-world industrial needs, SmartPilot empowers industries with intelligent decision-making and drives transformative innovation in manufacturing. The demonstration video, datasets, and supplementary materials are available at https://anonymous.4open.science/r/SmartPilot925MBI/README.md
Authors: Harleen Kaur Bagga, Jasmine Bernard, Sahil Shaheen, Sarthak Arora (2024)
Publication date: December 2024 - Preprint
Abstract: Sarcasm is hard to interpret as human beings. Being able to interpret sarcasm is often termed as a sign of intelligence, given the complex nature of sarcasm. Hence, this is a field of Natural Language Processing which is still complex for computers to decipher. This Survey performed delves into different aspects of sarcasm detection, to create an understanding of the underlying problems faced during detection, approaches used to solve this problem, and different forms of available datasets for sarcasm detection.
ML, Physics & Sustainability
Method and system for detecting and quantifying regeneration events in a diesel particulate filter
Inventors: Inventor 1, Harleen Kaur Bagga, Inventor 2, Bhushan D Patil, Inventor 3, Abhijit Vishwas Patil et al. (2024)
Publication date: February 2024
Application Number: US-18/590,783
Abstract: Methods and Systems are provided for detecting and quantifying regeneration events for diesel particulate filters in diesel engines. Parameters such as exhaust temperature, DPF differential pressure, and mass flow rates received from an engine management system are checked to determine if a regeneration is detected. If a regeneration event is detected, DPF differential pressures are logged and identified as pre-event differential pressures or post-event differential pressures. A soot burn quality value is determined using the pre-event and post-event differential pressures. The soot burn quality is used to score the soot burn and to generate alerts.
ML, Physics & Sustainability
Normal Operation Model Generation and Anomaly Detection for a Common Rail Internal Combustion Engine
Inventors: Inventor 4, Harleen Kaur Bagga (2025)
Publication date: January 2025
Application Number: US-18/966,864
Abstract: A method for generating a normal operation model indicative of a normal operational condition of a fuel rail of a common rail internal combustion engine. The method comprises receiving, from an engine management system of the common rail internal combustion engine via a data interface, engine model data obtained periodically during operation of the common rail internal combustion engine. The engine model data comprises fuel rail pressure data indicative of a fuel rail pressure of the fuel rail and engine speed data indicative of an engine speed of the engine. The method further comprises storing the engine model data in a memory. The engine model data is accumulated over a sampling time period. The method further comprises dividing, by a processor, the fuel rail pressure data into a predetermined number of engine speed bins derived from the engine speed data during the sampling time period so as to generate binned engine model data indicative of frequency of occurrence of fuel rail pressures within each bin, and generating, by the processor for each bin of the predetermined number of bins, statistical attribute model data based on the binned engine model data. The method also comprises generating, by the processor, the normal operation model based on the statistical attribute model data for each bin when taken together, and storing, in the memory, the normal operation model for comparison with engine test data accumulated over a detection time period for enabling detection of an anomaly in an operational condition of the fuel rail based on comparison between the engine test data accumulated over the detection time period and the statistical attribute model data of the normal operation model.
ML, Physics & Sustainability
Method and Model generation system for generating and oil quality monitoring model, and Method and detection system for detecting an anomaly in an oil pressure related operational model
Inventors: Inventor 5, Harleen Kaur Bagga (2025)
Publication date: In the process of filing
Abstract: A method for generating an oil quality monitoring model for an internal combustion engine of a vehicle. The method comprises receiving, from an engine management system of the internal combustion engine via a data interface, engine model data obtained periodically during a sequence of sampling time periods during operation of the internal combustion engine. The engine model data comprises oil pressure data indicative of oil pressure of the internal combustion engine, and engine speed data indicative of an engine speed of the engine. The method includes storing the engine model data in a memory for each sampling time period and dividing, by a processor, the oil pressure data into a predetermined number of engine speed bins derived from the engine speed data during each sampling time period so as to generate binned engine model data indicative of frequency of occurrence of oil pressures within each bin for each sampling time period. The method further includes generating, by the processor for each bin of the predetermined number of bins and for each sampling time period, statistical attribute model data based on the binned engine model data, and generating, by the processor, the oil quality monitoring model based on the statistical attribute model data for each bin when taken together for each sampling time period by labelling the statistical attribute model data based on operational data of the vehicle that coincides with the associated sampling time period to identify one or more suboptimal oil operation conditions. The method additionally includes storing, in the memory, the oil quality monitoring model for comparison with engine test data accumulated over detection time periods for enabling forecasting of engine oil anomalies of the internal combustion engine based on comparison between the engine test data for each sampling time period and the statistical attribute model data of the oil quality monitoring model.