Developed by AryaXAI, DL-Backtrace is a pioneering approach designed to redefine how we interpret and optimize deep learning systems. Going beyond typical explanations, DL-Backtrace offers a model-agnostic approach to dissecting the inner workings of neural networks, providing insights into how these models perceive data and make decisions. With DL-Backtrace, you can explain model outcomes in detail and measure each adjustment's impact, allowing for precision in fine-tuning, targeted improvements, and weight redistribution to align outcomes with business objectives.
In this session, Pratinav Seth, Research Scientist at AryaXAI, presents the core principles of DL-Backtrace and demonstrate its application in enhancing transparency for deep learning models across various use cases.
Pratinav Seth is a research scientist at AryaXAI, where he works on enhancing the reliability and interpretability of AI models. His expertise spans multiple AI domains, with a particular emphasis on creating resource-efficient models and promoting AI trustworthiness through explainability and uncertainty quantification. A recent graduate of the Manipal Institute of Technology, he was recognized as an AAAI Undergraduate Consortium Scholar in 2023. His research has been showcased at prestigious conferences, including CVPR and NeurIPS.
©2024 SSI Club. All Rights Reserved