Beyond Scaling: Exploring Alternative Pathways to AI Intelligence

Title of the Talk: Beyond Scaling: Exploring Alternative Pathways to AI Intelligence.
Speaker: Dr. Manoj Agarwal
Host Faculty: Dr.Sakethanath Jagarlapudi
Date: Jan 30, 2025
Time: 10am-11am
Venue: CS-LH3, EECS Building

Abstract:

Given a corpus of documents, it remains a hard question to answer “What is this data about?”. LLMs, although remarkable in generating seemingly intelligent answers, have no real understanding of the data, i.e., the semantic understanding of the concepts and meanings behind the words (though they “simulate” that understanding). Hence, these models present a few fundamental challenges for many critical use cases such as hallucination, even though their world knowledge is deep and growing. The other key problems are data leakage and context window limitation.

Besides, some other factors impacting the adoption of these models are latency and cost. Approaches, such as Retrieval Augmented Generation (RAG), are proposed to handle a few of these limitations. In this talk, we present a novel approach which is knowledge centric instead of document centric, to address the challenges outlined above.

Our first step is to semantically understand the data. Towards that, we build a knowledge layer that captures the ontology and the relationships between the entities and represents the data as a knowledge graph, and a semantic query engine to contextually parse the user query therefore enabling contextual search for complex queries. We propose a KG-RAG approach that grounds the small language model in the factual knowledge as well as improves its reasoning capabilities.

Biography

Dr. Manoj Agarwal is a co-founder of an artificial intelligence company, GiKA.AI, that aims to build better data intelligence systems for improved semantic search and analytics. Before this, he was Senior Staff Engineer in the Discovery intelligence team at Uber AI where he introduced the semantic search for Uber Eats. Prior to joining Uber, Manoj worked as Principal Applied Scientist at Microsoft – AI and Research and as a senior researcher in IBM Research. In Microsoft, Manoj was the chief architect for building a web scale product knowledge graph, comprising a few hundred million products. He also worked as adjunct faculty in IIT-Gandhinagar. Dr. Manoj Agarwal completed his PhD from IIT-Bombay. His work has been recognized with numerous awards including ACM India Doctoral Dissertation Award (Honorary Mention). His research interests are in the areas of web mining, graph mining, pattern recognition, data mining, knowledge graphs, language models and information retrieval with more than 30 patents and over 25 research papers.