A Low-cost Strategic Monitoring Approach for Scalable and Interpretable Error Detection in Deep Neural Networks
This addresses the need for scalable and interpretable error detection in deep computer vision networks, offering a low-cost solution for improving reliability and explainability, though it is incremental as it builds on existing anomaly detection techniques.
The paper tackles the problem of detecting silent data corruption in deep neural networks by introducing a compact run-time monitoring approach that uses strategically placed quantile markers on a few hidden layers, achieving up to ~96% precision and ~98% recall with minimal compute overhead (as low as 0.3%).
We present a highly compact run-time monitoring approach for deep computer vision networks that extracts selected knowledge from only a few (down to merely two) hidden layers, yet can efficiently detect silent data corruption originating from both hardware memory and input faults. Building on the insight that critical faults typically manifest as peak or bulk shifts in the activation distribution of the affected network layers, we use strategically placed quantile markers to make accurate estimates about the anomaly of the current inference as a whole. Importantly, the detector component itself is kept algorithmically transparent to render the categorization of regular and abnormal behavior interpretable to a human. Our technique achieves up to ~96% precision and ~98% recall of detection. Compared to state-of-the-art anomaly detection techniques, this approach requires minimal compute overhead (as little as 0.3% with respect to non-supervised inference time) and contributes to the explainability of the model.