# A method of limiting performance loss of CNNs in noisy environments

## TL;DR

This paper introduces a novel method to improve CNN robustness in noisy environments by dynamically adjusting neuron biases based on noise conditions, outperforming traditional approaches in accuracy and efficiency.

## Contribution

The paper presents a new bias adjustment technique for CNNs that enhances noise robustness without extensive retraining or preprocessing.

## Key findings

- The method improves recognition accuracy in noisy conditions.
- It reduces computational complexity compared to training multiple models.
- It outperforms existing denoising and specialized training approaches.

## Abstract

Convolutional Neural Network (CNN) recognition rates drop in the presence of noise. We demonstrate a novel method of counteracting this drop in recognition rate by adjusting the biases of the neurons in the convolutional layers according to the noise conditions encountered at runtime. We compare our technique to training one network for all possible noise levels, dehazing via preprocessing a signal with a denoising autoencoder, and training a network specifically for each noise level. Our system compares favorably in terms of robustness, computational complexity and recognition rate.