Biased Mixtures Of Experts: Enabling Computer Vision Inference Under Data Transfer Limitations
Preprint
Abstract
We propose a novel mixture-of-experts class to optimize computer vision models in accordance with data transfer limitations at test time. Our approach postulates that the minimum acceptable amount of data allowing for highly-accurate results can vary for different input space partitions. Therefore, we consider mixtures where experts require different amounts of data, and train a sparse gating function to divide the input space for each expert. By appropriate hyperparameter selection, our approach is able to bias mixtures of experts towards selecting specific experts over others. In this way, we show that the data transfer optimization between visual sensing and processing can be solved as a convex optimization problem.To demonstrate the relation between data availability and performance, we evaluate biased mixtures on a range of mainstream computer vision problems, namely: (i) single shot detection, (ii) image super resolution, and (iii) realtime video action classification. For all cases, and when experts constitute modified baselines to meet different limits on allowed data utility, biased mixtures significantly outperform previous work optimized to meet the same constraints on available data.Keywords
All Related Versions
This publication has 11 references indexed in Scilit:
- On the Integration of Optical Flow and Action Recognition2019
- Learning a Convolutional Neural Network for Image Compact-ResolutionIEEE Transactions on Image Processing, 2018
- Single-Image Super-Resolution Based on Rational Fractal InterpolationIEEE Transactions on Image Processing, 2018
- Real-Time Action Recognition With Deeply Transferred Motion Vector CNNsIEEE Transactions on Image Processing, 2018
- Video Salient Object Detection via Fully Convolutional NetworksIEEE Transactions on Image Processing, 2017
- Speed/Accuracy Trade-Offs for Modern Convolutional Object Detectors2017
- Low-Rank Embedding for Robust Image Feature ExtractionIEEE Transactions on Image Processing, 2017
- Layered Coding for Mobile Cloud Gaming Using Scalable Blinn-Phong LightingIEEE Transactions on Image Processing, 2016
- Microsoft COCO: Common Objects in Context2014
- Linear Optimization Problems with Inexact Data2006