Masked Vector Quantization Article Swipe

MAKE IMPACT

David D. Nguyen , David Leibowitz , ‪Surya Nepal‬ , Salil S. Kanhere · YOU? · · 2023 · Open Access · · DOI: https://doi.org/10.48550/arxiv.2301.06626

Generative models with discrete latent representations have recently demonstrated an impressive ability to learn complex high-dimensional data distributions. However, their performance relies on a long sequence of tokens per instance and a large number of codebook entries, resulting in long sampling times and considerable computation to fit the categorical posterior. To address these issues, we propose the Masked Vector Quantization (MVQ) framework which increases the representational capacity of each code vector by learning mask configurations via a stochastic winner-takes-all training regime called Multiple Hypothese Dropout (MH-Dropout). On ImageNet 64$\times$64, MVQ reduces FID in existing vector quantization architectures by up to $68\%$ at 2 tokens per instance and $57\%$ at 5 tokens. These improvements widen as codebook entries is reduced and allows for $7\textit{--}45\times$ speed-up in token sampling during inference. As an additional benefit, we find that smaller latent spaces lead to MVQ identifying transferable visual representations where multiple can be smoothly combined.

Type

preprint

Language

en

Landing Page

http://arxiv.org/abs/2301.06626

PDF

https://arxiv.org/pdf/2301.06626

OA Status

green

Related Works

10

OpenAlex ID

https://openalex.org/W4317462488

Raw OpenAlex JSON

OpenAlex ID

https://openalex.org/W4317462488

Canonical identifier for this work in OpenAlex

DOI

https://doi.org/10.48550/arxiv.2301.06626

Digital Object Identifier

Title

Masked Vector Quantization

Work title

Type

preprint

OpenAlex work type

Language

en

Primary language

Publication year

2023

Year of publication

Publication date

2023-01-16

Full publication date if available

Authors

David D. Nguyen, David Leibowitz, ‪Surya Nepal‬, Salil S. Kanhere

List of authors in order

Landing page

https://arxiv.org/abs/2301.06626

Publisher landing page

PDF URL

https://arxiv.org/pdf/2301.06626

Direct link to full text PDF

Open access

Yes

Whether a free full text is available

OA status

green

Open access status per OpenAlex

OA URL

https://arxiv.org/pdf/2301.06626

Direct OA link when available

Concepts

Codebook, Vector quantization, Categorical variable, Quantization (signal processing), Computer science, Inference, Dropout (neural networks), Learning vector quantization, Linde–Buzo–Gray algorithm, Computation, Security token, Leverage (statistics), Generative model, Generative grammar, Sampling (signal processing), Artificial intelligence, Algorithm, Sequence (biology), Pattern recognition (psychology), Machine learning, Computer vision, Computer security, Genetics, Biology, Filter (signal processing)

Top concepts (fields/topics) attached by OpenAlex

Cited by

0

Total citation count in OpenAlex

Related works (count)

10

Other works algorithmically related by OpenAlex

Full payload