arXiv subjects

Changyuan Liu

Publications and source records attributed to Changyuan Liu.

3 recordsLinked to original sources

arXiv2025

KD-MSLRT: Lightweight Sign Language Recognition Model Based on Mediapipe and 3D to 1D Knowledge Distillation

Artificial intelligence has achieved notable results in sign language recognition and translation. However, relatively few efforts have been made to significantly improve the quality of life for the 72 million hearing-impaired people worldwide. Sign language translation models, relying on video inputs, involves with large parameter sizes, making it time-consuming and computationally intensive to be deployed. This directly contributes to the scarcity of human-centered technology in this field. Additionally, the lack of datasets in sign language translation hampers research progress in this area. To address these, we first propose a cross-modal multi-knowledge distillation technique from 3D to 1D and a novel end-to-end pre-training text correction framework. Compared to other pre-trained models, our framework achieves significant advancements in correcting text output errors. Our model achieves a decrease in Word Error Rate (WER) of at least 1.4% on PHOENIX14 and PHOENIX14T datasets compared to the state-of-the-art CorrNet. Additionally, the TensorFlow Lite (TFLite) quantized model size is reduced to 12.93 MB, making it the smallest, fastest, and most accurate model to date. We have also collected and released extensive Chinese sign language datasets, and developed a specialized training vocabulary. To address the lack of research on data augmentation for landmark data, we have designed comparative experiments on various augmentation methods. Moreover, we performed a simulated deployment and prediction of our model on Intel platform CPUs and assessed the feasibility of deploying the model on other platforms.

cs.CY↗

arXiv2024

Study on the Particle Sorting Performance for Reactor Monte Carlo Neutron Transport on Apple Unified Memory GPUs

In simulation of nuclear reactor physics using the Monte Carlo neutron transport method on GPUs, the sorting of particles plays a significant role in performance of calculation. Traditionally, CPUs and GPUs are separated devices connected at low data transfer rate and high data transfer latency. Emerging computing chips tend to integrate CPUs and GPUs. One example is the Apple silicon chips with unified memory. Such unified memory chips have opened doors for new strategies of collaboration between CPUs and GPUs for Monte Carlo neutron transport. Sorting particle on CPU and transport on GPU is an example of such new strategy, which has been suffering the high CPU-GPU data transfer latency on the traditional devices with separated CPU and GPU. The finding is that for the Apple M2 max chip, sorting on CPU leads to better performance per power than sorting on GPU for the ExaSMR whole core benchmark problems and the HTR-10 high temperature gas reactor fuel pebble problem. The partially sorted particle order has been identified to contribute to the higher performance with CPU sort than GPU. The in-house code using both CPU and GPU achieves 7.5 times power efficiency that of OpenMC on CPU for ExaSMR whole core benchmark with depleted fuel, and 150 times for HTR-10 fuel pebble benchmark with depleted fuel.

cs.AR↗

arXiv2022

Monte Carlo neutron transport using low power mobile GPU devices

The using of GPU for Monte Carlo particle transport is lacking of fair comparisons. This work performs simulations on both CPU and GPU in the same package under the same manufacturing process of low power mobile devices. The experiment with simple pincell benchmark problems with fresh fuel gives consistent results between CPU and GPU. In the meanwhile, it finds that the Apple M1 GPU is as twice capable as M1 CPU, while entitled with a 5 times advantage in power consumption. The particle sorting algorithm optimized for GPU improves computing efficiency by 28\%, while prominently reducing GPU power consumption. Such advantage of sorting algorithm is expected to be greater for depleted fuel problems than fresh fuel problem. The kernel reconstruction Doppler broadening algorithm designed for continuously varying materials is demonstrated to produce consistent Doppler coefficients with the reference code and the algorithm can be efficiently implemented on GPU. Compared with the reference code with double precision floating point numbers, the testing codes with single precision floating point numbers could underestimate the K-effective values by about 500 pcm, and the Doppler coefficients of the fuel are well reproduced though. The conclusion may strengthen the argument that it is helpful for high performance computer to adopt GPU in order to reduce gross power consumption.

cs.DC↗