1k citations · 1k across the 6 of their papers we have counts for
7 papers
Benign Loss Landscapes Can Coexist with Worst-Case Hardness
Zach Furman, Stephan Wäldchen, Yangda Bei +1
Deep neural networks are expressive enough to contain worst-case targets that can be evaluated in polynomial time but cannot be learned in polynomial time by gradient descent. For…
Hardness of Deceptive Certificate Selection
Stephan Wäldchen
Recent progress towards theoretical interpretability guarantees for AI has been made with classifiers that are based on interactive proof systems. A prover selects a certificate fr…
Interpretability Guarantees with Merlin-Arthur Classifiers
Stephan Wäldchen, Kartikey Sharma, Berkant Turan +2
We propose an interactive multi-agent classifier that provides provable interpretability guarantees even for complex agents such as neural networks. These guarantees consist of low…
A Complete Characterisation of ReLU-Invariant Distributions
Jan Macdonald, Stephan Wäldchen
We give a complete characterisation of families of probability distributions that are invariant under the action of ReLU neural network layers. The need for such families arises du…
The Computational Complexity of Understanding Network Decisions
Stephan Wäldchen, Jan Macdonald, Sascha Hauch +1
For a Boolean function and an assignment to its variables we consider the problem of finding the subsets of the var…
A Rate-Distortion Framework for Explaining Neural Network Decisions
Jan Macdonald, Stephan Wäldchen, Sascha Hauch +1
We formalise the widespread idea of interpreting neural network decisions as an explicit optimisation problem in a rate-distortion framework. A set of input features is deemed rele…
Unmasking Clever Hans Predictors and Assessing What Machines Really Learn
Sebastian Lapuschkin, Stephan Wäldchen, Alexander Binder +3
Current learning machines have successfully solved hard application problems, reaching high accuracy and displaying seemingly "intelligent" behavior. Here we apply recent technique…