A stepped sampling method for video detection using LSTM

TL;DR

This paper introduces a stepped sampling method for LSTM-based video detection that enhances temporal information fusion, leading to faster convergence and higher stability in training, inspired by human memory processes.

Contribution

The paper proposes a novel stepped sampler based on repeated input for LSTM, improving training efficiency and accuracy over traditional sampling methods.

Findings

Faster convergence of training loss with the stepped sampler

More stable training loss after convergence

Higher test accuracy compared to traditional samplers

Abstract

Artificial neural networks that simulate human achieves great successes. From the perspective of simulating human memory method, we propose a stepped sampler based on the "repeated input". We repeatedly inputted data to the LSTM model stepwise in a batch. The stepped sampler is used to strengthen the ability of fusing the temporal information in LSTM. We tested the stepped sampler on the LSTM built-in in PyTorch. Compared with the traditional sampler of PyTorch, such as sequential sampler, batch sampler, the training loss of the proposed stepped sampler converges faster in the training of the model, and the training loss after convergence is more stable. Meanwhile, it can maintain a higher test accuracy. We quantified the algorithm of the stepped sampler. We assume that, the artificial neural networks have human-like characteristics, and human learning method could be used for machine learning.