About
Research & Development
Spiking Neural Networks Offer Energy-Efficient AI for Edge Devices in Time Series Analysis
TLDR: This research introduces a general framework for training Spiking Neural Networks (SNNs) for binary classification in multivariate time series, focusing on energy efficiency...
Research & Development
Self-Rewarding PPO: Improving LLM Generalization from Demonstrations
TLDR: Self-Rewarding PPO (SRPPO) is a novel method for aligning large language models (LLMs) that combines supervised fine-tuning (SFT) with proximal policy optimization (PPO)....
Research & Development
Navigating Uncertainty: A New AI Framework for Complex Decision-Making
TLDR: ESCORT is a new particle-based AI framework that improves how agents represent and update their understanding of uncertain environments (Partially Observable Markov Decision...
Research & Development
RePULSe: A New Method to Drastically Reduce Undesirable Language Model Outputs
TLDR: RePULSe is a novel training method for language models that significantly reduces the probability of undesirable outputs while maintaining high average performance. It...
Research & Development
Optimizing AI Model Learning with a Dynamic Gompertz Curve Approach
TLDR: This paper introduces Gompertz-CNN, a novel dynamic knowledge distillation framework that uses the Gompertz growth model to adaptively adjust distillation loss weights during...
Research & Development
Improving AI Decision-Making by Tackling Unseen Factors
TLDR: This paper introduces Causal DQN, a novel deep reinforcement learning algorithm designed to learn effective policies even when data is biased by unobserved...