A5103083219- Logic, Reasoning, and Knowledge
- Computational Drug Discovery Methods
- Data Mining Algorithms and Applications
- Machine Learning and Algorithms
- Topic Modeling
- Semantic Web and Ontologies
- Advanced Algebra and Logic
- Natural Language Processing Techniques
- Explainable Artificial Intelligence (XAI)
- Rough Sets and Fuzzy Logic
- Machine Learning and Data Classification
- Bayesian Modeling and Causal Inference
- Neural Networks and Applications
- Biomedical Text Mining and Ontologies
- COVID-19 diagnosis using AI
- Logic, programming, and type systems
- Machine Learning in Materials Science
- Analytical Chemistry and Chromatography
- Machine Learning in Bioinformatics
- Domain Adaptation and Few-Shot Learning
- AI-based Problem Solving and Planning
- Pharmacogenetics and Drug Metabolism
- Handwritten Text Recognition Techniques
- Advanced Database Systems and Queries
- Data Stream Mining Techniques
Birla Institute of Technology and Science, Pilani
2016-2025
Virginia Tech
2025
Birla Institute of Technology and Science, Pilani - Goa Campus
2016-2024
Velammal Medical College Hospital and Research Institute
2023-2024
UNSW Sydney
2007-2022
Macquarie University
2022
Microsoft Research (United Kingdom)
2022
University of Trinidad and Tobago
2021
Carnegie Mellon University
2004-2019
Association for Computing Machinery
2019
We present a general approach to forming structure-activity relationships (SARs). This is based on representing chemical structure by atoms and their bond connectivities in combination with the inductive logic programming (ILP) algorithm PROGOL. Existing SAR methods describe using attributes which are properties of an object. It not possible map directly attribute-based descriptions, as such descriptions have no internal organization. A more natural way use relational description, where...
Abstract Summary: We initiated the Predictive Toxicology Challenge (PTC) to stimulate development of advanced SAR techniques for predictive toxicology models. The goal this challenge is predict rodent carcinogenicity new compounds based on experimental results US National Program (NTP). Submissions will be evaluated quantitative and qualitative scales select most models those with highest toxicological relevance. Availability: http://www.informatik.uni-freiburg.de/~ml/ptc/ Contact:...
Inductive Logic Programming (ILP) is an area of Machine Learning which has now reached its twentieth year. Using the analogy a human biography this paper recalls development subject from infancy through childhood and teenage years. We show how in each phase ILP been characterised by attempt to extend theory implementations tandem with novel challenging real-world applications. Lastly, projection we suggest directions for research will help coming age.
Abstract Motivation: The development of in silico models to predict chemical carcinogenesis from molecular structure would help greatly prevent environmentally caused cancers. Predictive Toxicology Challenge (PTC) competition was organized test the state-of-the-art applying machine learning form such predictive models. Results: Fourteen groups generated 111 use Receiver Operating Characteristic (ROC) space allowed be uniformly compared regardless error cost function. We developed a...
We introduce a novel algorithm for decision tree learning in the multi-instance setting as originally defined by Dietterich et al. It differs from existing learners few crucial, well-motivated details. Experiments on synthetic and real-life datasets confirm beneficial effect of these differences show that resulting system outperforms learners.
The machine learning program Progol was applied to the problem of forming structure-activity relationship (SAR) for a set compounds tested carcinogenicity in rodent bioassays by U.S. National Toxicology Program (NTP). is first inductive logic programming (ILP) algorithm use fully relational method describing chemical structure SARs, based on using atoms and their bond connectivities. well suited SARs as it designed produce easily understandable rules (structural alerts) sets noncongeneric...
In the recent years, smart sensing approach is creating a vibrant impact in shaping our future. The growth of technology can be incorporated with rising events triggering need for better lifestyle. Recent technological advancement which has influenced change lifestyle, field IoT (Internet Things). applications are vast and innumerable. One such application implied Automobile industry to improve quality safety vehicles. With increase number accidents deteriorating performance cars, there...
Dense retrieval (DR) methods conduct text by first encoding texts in the embedding space and then matching them nearest neighbor search. This requires strong locality properties from representation space, e.g., close allocations of each small group relevant texts, which are hard to generalize domains without sufficient training data. In this paper, we aim improve generalization ability DR models source with rich supervision signals target any relevance label, zero-shot setting. To achieve...
Our interest is in the generation of "lead" molecules early-stage drug design. Leads are small (ligands) that can bind to a part pre-specified target and also satisfy multiple physico-chemical constraints. We propose using techniques developed Inductive Logic Programming (ILP) identify logical specification feasible molecules; then this construct program uses large language model (LLM) generate new molecules. ensure constructed correct, sense every molecule generated by according...