W2WNet: A two-module probabilistic Convolutional Neural Network with embedded data cleansing functionality

Ideally, Convolutional Neural Networks (CNNs) should be trained with high quality images with minimum noise and correct ground truth labels. Nonetheless, in many real-world scenarios, such high quality is very hard to obtain, and datasets may be affected by any sort of image degradation and mislabelling issues. This negatively impacts the performance of standard CNNs, both during the training and the inference phase. To address this issue we propose Wise2WipedNet (W2WNet), a new two-module Convolutional Neural Network, where a Wise module exploits Bayesian inference to identify and discard spurious images during the training and a Wiped module takes care of the final classification, while broadcasting information on the prediction confidence at inference time. The goodness of our solution is demonstrated on a number of public benchmarks addressing different image classification tasks, as well as on a real-world case study on histological image analysis. Overall, our experiments demonstrate that W2WNet is able to identify image degradation and mislabelling issues both at training and at inference time, with positive impact on the final classification accuracy.

Automated summary

In real-world scenarios, training Convolutional Neural Networks (CNNs) with high quality images and correct labels is difficult. This affects the performance of CNNs during both training and inference. To tackle this issue, we propose a new two-module CNN called Wise2WipedNet (W2WNet), which uses Bayesian inference to identify and discard spurious images during training and provides prediction confidence during inference. Our experiments on various image classification tasks and histological image analysis demonstrate that W2WNet can effectively identify image degradation and mislabelling issues, resulting in improved classification accuracy.