Embedded Files
Methods and Systems for the Automated Generation of Neural Network Architectures
Methods and Systems for the Automated Generation of Neural Network Architectures
U.S. Provisional Patent Application No. 63/627,621
Inventors: Md Hafizur Rahman, Prabuddha Chakraborty & Zafaryab Haider.
Description: This framework designed to automatically generate architecture with user preferences like FPS, Accuracy, Energy etc. [See Full Paper]
Methods and Systems for the Design and Use of Intelligent Layer Shared Neural Networks and Model Architecture Search Algorithms
Methods and Systems for the Design and Use of Intelligent Layer Shared Neural Networks and Model Architecture Search Algorithms
U.S. Provisional Patent Application No. 63/535,627
Inventors: Prabuddha Chakraborty, Md Hafizur Rahman, Sumaiya Shomaji & Md Mashfiq Rizvee.
Description: This framework designed to search network for multi-task applications with layer sharing approach. Its predictive approach reduces the overall searching time. [See Full Paper]
Leakage-Aware Secure Transformations (LAST) and POLARIS for Power Side-Channel Mitigation
Leakage-Aware Secure Transformations (LAST) and POLARIS for Power Side-Channel Mitigation
U.S. Provisional Patent Application No. 63/827,407
Inventors: Prabuddha Chakraborty, Tanzim Mahfuz, Tasneem Farhana Suha, Swarup Bhunia, Sudipta Paria
Description: A novel AI-driven framework that generates and optimally inserts masked logic solutions to significantly reduce power side-channel leakage in digital hardware. [See Full Paper]
Framework for Mitigating Malicious RLHF Feedback in LLM Training Using Consensus Based Reward
Framework for Mitigating Malicious RLHF Feedback in LLM Training Using Consensus Based Reward
U.S. Provisional Patent Application No. 63/768,361
Inventors: Zafaryab Haider, Md Hafizur Rahman, Shane Moeykens & Prabuddha Chakraborty.
Description: This framework makes AI training safer, especially when you can’t fully control or verify every human giving feedback. (Full Paper)
Live Tracking of Cyber Components and AI Guided Automated Vulnerability Detection & Fixing
Live Tracking of Cyber Components and AI Guided Automated Vulnerability Detection & Fixing
U.S. Provisional Patent Application No.
Inventors: Nickolas Millett, Katherine Arsenault, and Prabuddha Chakraborty.
Description: This framework tracks cyber components in real time for use in creating a digital twin. This digital twin can then be used to conduct automated vulnerability assessments using AI which fixes can then be applied to the production environment.
Page updated
Google Sites
Report abuse