Tag: 2024
All the talks with the tag "2024".
Scalable MatMul-free Language Modeling
Sagar Prakash BaradPublished: at 02:00 PMThis talk presents a paper that proposes a scalable MatMul-free language model, challenging the assumption that matrix multiplications are essential for high-performing language models. The paper demonstrates that by using ternary weights and element-wise Hadamard products, MatMul operations can be completely removed from large language models while maintaining strong performance. The paper provides an optimized implementation of the MatMul-free language model, achieving significant reductions in memory usage and latency compared to conventional models.
SplaTAM - Splat, Track & Map 3D Gaussians for Dense RGB-D SLAM
Anubhav VishwakarmaPublished: at 02:00 PMThis paper introduces SplaTAM, an approach that leverages explicit volumetric representations, specifically 3D Gaussians, for dense simultaneous localization and mapping (SLAM) from a single RGB-D camera. SplaTAM achieves high-fidelity reconstruction and superior performance in camera pose estimation, map construction, and novel-view synthesis compared to existing methods. The implementation code for SplaTAM is available on GitHub.
AI in Cybersecurity
Divy AgnihotriPublished: at 02:00 PMThis paper presents an overview of AI-driven cybersecurity, highlighting the use of machine learning, deep learning, natural language processing, and knowledge-based expert systems. It discusses the role of AI in protecting Internet-connected systems from cyber threats and attacks. The paper provides insights into the security intelligence modeling based on AI methods and identifies future research directions in the field.
Quantum Machine Learning — An Overview
Kirtidev ParidaPublished: at 02:00 PMThis paper provides an overview of quantum machine learning (QML) and its potential for computational acceleration. It discusses the current state of QML, benchmarking the performance of classical and quantum algorithms on various tasks. The paper highlights the use of hybrid methods and novel approaches in QML, demonstrating their promise in predicting quantum states and enhancing classical data science algorithms.
Application of a New Machine Learning Model to Improve Earthquake Ground Motion Predictions
Dibyanshu MohapatraPublished: at 02:00 PMThis paper presents a cross-region prediction model named SeisEML for predicting peak ground acceleration (PGA) during earthquakes. The SeisEML model combines hybridized models, kernel-based algorithms, tree regression algorithms, and regression algorithms to achieve improved accuracy compared to conventional attenuation relations. The model has been tested on datasets from Japan and Iran, demonstrating its potential for regional and global earthquake predictions.
LightGlue - Local Feature Matching at Light Speed
Adyasha M.Published: at 02:00 PMThis paper introduces LightGlue, a deep neural network that learns to match local features across images. It presents improvements over the state-of-the-art sparse matching method, SuperGlue, making LightGlue more efficient, accurate, and easier to train. The paper provides the implementation code for LightGlue.