Deep k-Nearest Neighbors: Towards Confident, Interpretable and Robust Deep Learning
This addresses robustness and interpretability issues in deep learning for applications such as image recognition and malware detection, but it is an incremental step as it combines existing methods.
The paper tackles the lack of robustness and interpretability in deep neural networks by introducing Deep k-Nearest Neighbors (DkNN), a hybrid classifier that uses k-nearest neighbors on DNN layer representations to provide confidence estimates and human-interpretable explanations, showing it accurately identifies out-of-distribution inputs like adversarial examples.
Deep neural networks (DNNs) enable innovative applications of machine learning like image recognition, machine translation, or malware detection. However, deep learning is often criticized for its lack of robustness in adversarial settings (e.g., vulnerability to adversarial inputs) and general inability to rationalize its predictions. In this work, we exploit the structure of deep learning to enable new learning-based inference and decision strategies that achieve desirable properties such as robustness and interpretability. We take a first step in this direction and introduce the Deep k-Nearest Neighbors (DkNN). This hybrid classifier combines the k-nearest neighbors algorithm with representations of the data learned by each layer of the DNN: a test input is compared to its neighboring training points according to the distance that separates them in the representations. We show the labels of these neighboring points afford confidence estimates for inputs outside the model's training manifold, including on malicious inputs like adversarial examples--and therein provides protections against inputs that are outside the models understanding. This is because the nearest neighbors can be used to estimate the nonconformity of, i.e., the lack of support for, a prediction in the training data. The neighbors also constitute human-interpretable explanations of predictions. We evaluate the DkNN algorithm on several datasets, and show the confidence estimates accurately identify inputs outside the model, and that the explanations provided by nearest neighbors are intuitive and useful in understanding model failures.