Huzi Cheng

The paper introduces a biologically plausible variant of backpropagation through time using replay in a reversible recurrent neural network, demonstrating its effectiveness in episodic learning.

This study analyzes AI-generated text infiltration in scientific preprints using LLM detectors, finding widespread use across domains and positive impacts on citation numbers.

This study introduces a novel RNN model called SPEL that mimics prefrontal cortex activity by minimizing sequence prediction error, showing it matches neural responses and behavior in monkeys.

This paper introduces a goal reduction mechanism that accelerates reinforcement learning and models human brain activity in goal-directed learning by deriving subgoals through loop-removal.

This paper develops a recurrent neural network model for decision making that mimics animal behavior and neural activity patterns in tasks involving sensory input, action, and reward.