Hidden Monotonicity: Explaining Deep Neural Networks via their DC Decomposition
By: Jakob Paul Zimmermann, Georg Loho
Potential Business Impact:
Makes AI show why it makes decisions.
It has been demonstrated in various contexts that monotonicity leads to better explainability in neural networks. However, not every function can be well approximated by a monotone neural network. We demonstrate that monotonicity can still be used in two ways to boost explainability. First, we use an adaptation of the decomposition of a trained ReLU network into two monotone and convex parts, thereby overcoming numerical obstacles from an inherent blowup of the weights in this procedure. Our proposed saliency methods -- SplitCAM and SplitLRP -- improve on state of the art results on both VGG16 and Resnet18 networks on ImageNet-S across all Quantus saliency metric categories. Second, we exhibit that training a model as the difference between two monotone neural networks results in a system with strong self-explainability properties.
Similar Papers
Explaining Deep Network Classification of Matrices: A Case Study on Monotonicity
Machine Learning (CS)
Finds simple math rules for sorting numbers.
Logical Expressivity and Explanations for Monotonic GNNs with Scoring Functions
Machine Learning (CS)
Explains computer predictions by finding simple rules.
On the Depth of Monotone ReLU Neural Networks and ICNNs
Machine Learning (CS)
Makes computers better at finding the biggest number.