Explaining A Black-box By Using A Deep Variational Information Bottleneck Approach

Abstract

Interpretable machine learning has gained much attention recently. Briefness and comprehensiveness are necessary in order to provide a large amount of information concisely when explaining a black-box decision system. However, existing interpretable machine learning methods fail to consider briefness and comprehensiveness simultaneously, leading to redundant explanations. We propose the variational information bottleneck for interpretation, VIBI, a system-agnostic interpretable method that provides a brief but comprehensive explanation. VIBI adopts an information theoretic principle, information bottleneck principle, as a criterion for finding such explanations. For each instance, VIBI selects key features that are maximally compressed about an input (briefness), and informative about a decision made by a black-box system on that input (comprehensive). We evaluate VIBI on three datasets and compare with state-of-the-art interpretable machine learning methods in terms of both interpretability and fidelity evaluated by human and quantitative metrics.

Dual-interest Factorization-heads Attention for Sequential Recommendation

“…From this figure, we can observe that: (1) For the collected Micro-video dataset, users are still willing to watch videos even after they receive the disliked videos. This may be because the negative recommended videos are of low cost for users as they can easily skip the disliked videos, making no significant impact on their later preferred videos; (2) For the e-commerce dataset about Amazon, we can discover that when the source feedback is negative, the probability of target feedback being negative will increase sharply. This may be because the negative purchased items are of high cost in e-commerce for users as it will waste their money, increasing their unsatisfied emotion sharply.…”

“…For example, L2X [3] exploits Gumbel-softmax [13] for feature selection by instance, with its hard attention design [39]. Moreover, VIBI [2] further propose a feature score constraint in a global prior so as to simplify and purify the explainable representation learning. As self-attention is popular [6,28], there is also a work that explains what heads learn and concludes that some redundant heads can be pruned [30].…”

“…• Comparing with Multi-heads Attention: our factorizationheads attention incorporates the interaction between different heads with additional four lines at lines 12-14 and 17, which are transpose and reshape operations and with only 𝑂 (1) temporal complexity. 2 . • Comparing with Talking-heads Attention: our factorizationheads attention achieves explicit interaction with additional transpose and reshape operations at 𝑂 (1) temporal complexity while talking-heads attention achieves implicit interaction with two matrix multiplication operations at 𝑂 (𝑚 ×ℎ 𝑘 ×ℎ) and 𝑂 (𝑚 ×ℎ ×ℎ 𝑣 ) temporal complexities 3 , respectively.…”

Listen to Interpret: Post-hoc Interpretability for Audio Networks with NMF

“…The most suitable baselines to benchmark its fidelity are post-hoc methods that approximate the classifier over input space with a single surrogate model. We select two state-of-the-art systems, FLINT [23] and VIBI [58]. A variant of our own proposed method, L2I w/ Θ MAX , is also evaluated.…”

Knowledge Mining: A Cross-disciplinary Survey

“…While most of the recent efforts in interpretability are focused on developing new methods for computing local explanations (e.g., using game theory theorems [129] , information theory principles [130] , or a case-based reasoning approach [131] ), one of the biggest challenges is evaluating both the faithfulness of extracted knowledge and the usefulness of such knowledge for humans to understand the black-box model′s decision process [132] .…”