Yiming Cheng

Yiming Cheng


Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
21 March 2025 in Applied Sciences
Verified
Journal Article
Open access
PDF file
运谢运泽 谢
运泽 谢
YMYiping Ma
Yiping Ma
YCYiming Cheng
Yiming Cheng
ZLZhiqian Li
Zhiqian Li
XLXiaoyu Liu
Xiaoyu Liu
This research proposes a hybrid path planning framework for intelligent inland waterway Unmanned Surface Vehicles (USVs), which integrates the enhanced BIT* (Batch Informed Trees) algorithm with the TD3 (Twin Delayed Deep Deterministic Policy Gradient) deep reinforcement learning method. To address the limitations of traditional path planning algorithms in dynamic environments, the proposed BIT*+TD3 model leverages the BIT* algorithm to generate initial paths in static environments through elliptical informed sampling and heuristic search. Simultaneously, it utilizes the TD3 algorithm to dynamically optimize these paths through twin Critic networks and delayed policy updates. This research designs a novel reward mechanism aimed at minimizing turning angles, smoothing speed transitions, and shortening path lengths. Furthermore, it incorporates a hydrodynamics-based energy consumption model and multi-threaded parallel computation to enhance computational efficiency. Experimental validation demonstrates that, compared to traditional methods, this model exhibits significant improvements in obstacle avoidance success rate, safe distance maintenance, convergence speed, and smoothness. By bridging sampling-based planning methods with deep reinforcement learning methods, this research advances autonomous navigation technology and provides a scalable and energy-efficient solution for maritime applications.

15 May 2025 in arXiv
Verified
Preprint
Open access
PDF file
JSJintian Shao
Jintian Shao
HHHongyi Huang
Hongyi Huang
JWJiayi Wu
Jiayi Wu
YCYiming Cheng
Yiming Cheng
ZWZhiYu Wu
ZhiYu Wu
MZMingKai Zheng
MingKai Zheng
Large Language Models (LLMs) have achieved remarkable success but face significant computational and memory challenges, particularly due to their extensive output vocabularies. The final linear projection layer, mapping hidden states to vocabulary-sized logits, often constitutes a substantial portion of the model's parameters and computational cost during inference. Existing methods like adaptive softmax or hierarchical softmax introduce structural complexities. In this paper, we propose VQ-Logits, a novel approach that leverages Vector Quantization (VQ) to drastically reduce the parameter count and computational load of the LLM output layer. VQ-Logits replaces the large V * dmodel output embedding matrix with a small, shared codebook of K embedding vectors (K << V ). Each token in the vocabulary is mapped to one of these K codebook vectors. The LLM predicts logits over this compact codebook, which are then efficiently "scattered" to the full vocabulary space using the learned or preassigned mapping. We demonstrate through extensive experiments on standard language modeling benchmarks (e.g., WikiText-103, C4) that VQ-Logits can achieve up to 99% parameter reduction in the output layer and 6x speedup in logit computation, with only a marginal 4% increase in perplexity compared to full softmax baselines. We further provide detailed ablation studies on codebook size, initialization, and learning strategies, showcasing the robustness and effectiveness of our approach.