Subhankar Mishra Lab Weekly Talks
RSS FeedOn this site, you can find the files of the talks & presentations by our lab members and interns on various Machine Learning or Computer Science topics.
Feel free to reach out to us on any of our socials below.
Recent Talks
Large Concept Models
Adhilsha AnsadPublished: at 07:00 PMThis talk will will explore the training objectives, segmentation techniques, and generation strategies of Large Concept Models (LCMs). LCMs are a novel class of models that leverage sentence-level tokenization to represent concepts, a higher abstraction than current tokens. We will also discuss the quantization of LCMs and their potential applications in various domains.
Quantum Implicit Neural Representations
Haribandhu JenaPublished: at 07:00 PMThis short talk discusses QIREN. QIREN is a quantum generalization of Fourier Neural Networks (FNNs) for implicit neural representations, offering a quantum advantage in modeling high-frequency signal components. It outperforms SoTA models in signal representation, image superresolution, and image generation.
The Expressive Power of GNNs
Pankaj KumarPublished: at 02:00 PMThis talk delves into the theoretical foundations of Graph Neural Networks (GNNs), focusing on their ability to model complex relationships between node features and graph structures. While GNNs excel in tasks like node classification and link prediction, their expressive power remains challenging to quantify due to the interplay of graph topology and node features. The discussion highlights key challenges, such as trade-offs between expressiveness and computational efficiency, and explores advancements like the connection between GNNs and the Weisfeiler-Lehman graph isomorphism test. The talk also outlines future directions for developing rigorous evaluation methods to better understand and assess GNN expressivity.
Private Graph Extraction via Feature Explanation
Rishi Raj SahooPublished: at 02:00 PMThis study explores privacy leakage in GNNs through graph reconstruction attacks using feature explanations. It compares explanation methods—gradient-based, perturbation-based, and surrogate model-based—showing that explanations aid graph reconstruction, with a trade-off between privacy and utility. A defense using randomized response is proposed to reduce attack success.
Spiking Neural Network
Sagar Prakash BaradPublished: at 07:00 PMExploring spiking neural networks and their role in modeling memory storage in the brain.