Distill-VQ: Learning Retrieval Oriented Vector Quantization By Distilling Knowledge from Dense Embeddings.
Shitao XiaoZheng LiuWeihao HanJianjin ZhangDefu LianYeyun GongQi ChenFan YangHao SunYingxia ShaoXing XiePublished in: SIGIR (2022)
Keyphrases
- vector quantization
- image compression
- vector quantizer
- reduced complexity
- distortion measure
- successive approximation
- finite state vector quantization
- wavelet transform
- vector quantized
- index table
- fractal image compression
- input vector
- speaker recognition
- entropy constrained
- entropy constrained vector quantization
- neural network
- fractal image coding
- neural gas
- image representation
- unsupervised learning
- image retrieval
- image processing