si4onnx: A Python package for Selective Inference in Deep Learning Models
This addresses the challenge of establishing reliability in AI systems for researchers and practitioners using deep learning, though it is incremental as it builds on existing selective inference techniques.
The authors tackled the problem of evaluating the statistical significance of regions identified by deep learning methods like CAM and VAE-based anomaly detection, which may lack meaningful significance, by introducing si4onnx, a Python package that enables hypothesis testing with controlled type I error rates for selective inference in deep learning models.
In this paper, we introduce si4onnx, a package for performing selective inference on deep learning models. Techniques such as CAM in XAI and reconstruction-based anomaly detection using VAE can be interpreted as methods for identifying significant regions within input images. However, the identified regions may not always carry meaningful significance. Therefore, evaluating the statistical significance of these regions represents a crucial challenge in establishing the reliability of AI systems. si4onnx is a Python package that enables straightforward implementation of hypothesis testing with controlled type I error rates through selective inference. It is compatible with deep learning models constructed using common frameworks such as PyTorch and TensorFlow.