Stochastic Resonance Improves the Detection of Low Contrast Images in Deep Learning Models
This addresses the challenge of improving image classification under poor visibility conditions, but it is incremental as it applies a known phenomenon to a new context.
The study tackled the problem of detecting low-contrast images in deep learning models by adding controlled noise during testing, which partially recovered classification performance in an LSTM network for digit recognition.
Stochastic resonance describes the utility of noise in improving the detectability of weak signals in certain types of systems. It has been observed widely in natural and engineered settings, but its utility in image classification with rate-based neural networks has not been studied extensively. In this analysis a simple LSTM recurrent neural network is trained for digit recognition and classification. During the test phase, image contrast is reduced to a point where the model fails to recognize the presence of a stimulus. Controlled noise is added to partially recover classification performance. The results indicate the presence of stochastic resonance in rate-based recurrent neural networks.