Compressing Deep Convolutional Networks using Vector Quantization
- 类型:arxiv
- 标识:1412.6115
- 链接:https://arxiv.org/abs/1412.6115
- 主题:database
- 主分类:llm-infra
- 形态:method
- 被引:1235
- 被引来源:Semantic Scholar
- S2被引:1235
- OpenAlex被引:1019
- 影响力被引:46
- TLDR:This paper is able to achieve 16-24 times compression of the network with only 1% loss of classification accuracy using the state-of-the-art CNN, and finds in terms of compressing the most storage demanding dense connected layers, vector quantization methods have a clear gain over existing matrix factorization methods.
- OpenAlex ID:W1724438581
- OpenAlex DOI:10.48550/arxiv.1412.6115
- DOI:10.48550/arxiv.1412.6115
- DOI来源:OpenAlex
- 开放获取:green
- 开放获取链接:https://arxiv.org/pdf/1412.6115
- OpenAlex更新:2026-08-19
- 待LLM分类:否
- 标题中文:使用向量量化压缩深度卷积网络
- TLDR中文:本文在使用 SOTA CNN 的情况下,实现了 16–24 倍的网络压缩,仅带来 1% 的分类准确率损失,并发现针对存储开销最大的全连接层进行压缩时,向量量化方法相比现有矩阵分解方法具有明显优势。
- 来源文件:
- [OpenAlex discover]
- [S2 enrich]