Balaji Ganesan
Balaji Ganesan
  • Видео 10
  • Просмотров 682
Knowledge Enhanced Reasoning in LLM Agents
LLM Agents MOOC Hackathon Submission
Team: Web Weavers
Members: Balaji, Satyam, Laura, Rosni, Shetu, Ashvanth
Recent advances in Large Language Models (LLMs) have led to their widespread adoption for a variety of tasks, including question answering, summarization, and text-to-SQL conversion. However, LLMs often struggle when queries require external knowledge and complex reasoning beyond the original training corpus. Inspired by the Sherpas framework, we propose a Crew AI-based agentic solution that enhances reasoning capabilities by integrating external knowledge sources and structured interactions. Specifically, we focus on improving Text-to-SQL performance by leveraging a multi-agent arch...
Просмотров: 16

Видео

Explainable AI for Master Data Management
Просмотров 182 года назад
Sponsors' talk at CODS COMAD 2022.
Automated Evaluation of GNN Explanations with Neuro-Symbolic Reasoning
Просмотров 843 года назад
NeurIPS 2021 Demo. Vanya BK, Muhammed Ameen, Balaji Ganesan, Devbrat Sharma, Arvind Agarwal.
Case Similarity Demo
Просмотров 203 года назад
A demo of the case similarity task on a case graph from courts in India. We began by constructing a legal knowledge graph and then using the entities and relations for learning representation in a case graph.
Explainable Entity Matching
Просмотров 1803 года назад
We present 3 real world entity matching scenarios, entity resolution, large entity explanation and spatio-temporal search for related entities.
Regulatory Q&A System
Просмотров 74 года назад
Anu is a regulatory Q&A system that combines search index and a knowledge graph to answer questions on regulatory documents.
Explainable Link Prediction for Privacy Preserving Contact Tracing
Просмотров 164 года назад
Balaji Ganesan, Hima Patel, Sameep Mehta. Explainable Link Prediction for Privacy Preserving Contact Tracing. Presented as a poster at the International Workshop on Scalability, Privacy and Security in Federated Learning (SpicyFL 2020) co-located at NeurIPS 2020.
Anu Question Answering System
Просмотров 404 года назад
ISWC 2020 Industry Track presentation. Paper 513. Anu Question Answering System. Balaji Ganesan, Avirup Saha, Jaydeep Sen, Matheen Ahmed Pasha, Sumit Bhatia, Arvind Agarwal.
Explainable Link Prediction
Просмотров 2464 года назад
Learning graph representations from n-ary relational data and predicting links between entities has a number of real world applications in anti-money laundering, fraud detection, sales recommendations, etc. In this demo, we present our latest research to predict links between entities in TACRED dataset (as an example use case) using graph neural networks along with human understandable explanat...
Cognitive Compliance
Просмотров 557 лет назад
A short video describing my work. Made as part of a leadership training exercise.