DecTrain: Deciding When to Train a DNN Online

2024-10-03

arXiv.org

AI Summary

Monocular depth estimation models suffer from accuracy degradation on out-of-distribution data and prohibitively high computational overhead under continuous online training. To address this, we propose DecTrain—a sparse online training framework that dynamically triggers lightweight self-supervised fine-tuning based on predictive cost-benefit analysis of accuracy gain versus training cost. DecTrain is the first method enabling compact models to approach the generalization performance of large models through intelligent training scheduling. Experiments demonstrate that DecTrain achieves comparable accuracy to full-time online training while reducing training frequency by 44%. When applied to a small model, DecTrain recovers 97% of the accuracy attainable by a large baseline model, with overall computation reduced by 56% relative to that large model.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Computer Vision: Large Vision Models
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Deep neural networks (DNNs) can deteriorate in accuracy when deployment data differs from training data. While performing online training at all timesteps can improve accuracy, it is computationally expensive. We propose DecTrain, a new algorithm that decides when to train a monocular depth DNN online using self-supervision with low overhead. To make the decision at each timestep, DecTrain compares the cost of training with the predicted accuracy gain. We evaluate DecTrain on out-of-distribution data, and find DecTrain maintains accuracy compared to online training at all timesteps, while training only 44% of the time on average. We also compare the recovery of a low inference cost DNN using DecTrain and a more generalizable high inference cost DNN on various sequences. DecTrain recovers the majority (97%) of the accuracy gain of online training at all timesteps while reducing computation compared to the high inference cost DNN which recovers only 66%. With an even smaller DNN, we achieve 89% recovery while reducing computation by 56%. DecTrain enables low-cost online training for a smaller DNN to have competitive accuracy with a larger, more generalizable DNN at a lower overall computational cost.

Problem

• Monocular Depth Perception
• Online Learning
• Computational Cost

Innovation

• DecTrain
• Online Learning
• Computational Efficiency