FastIF: Scalable Influence Functions for Efficient Model Interpretation and Debugging

Abstract

Influence functions approximate the "influences" of training data-points for test predictions and have a wide variety of applications. Despite the popularity, their computational cost does not scale well with model and training data size. We present FastIF, a set of simple modifications to influence functions that significantly improves their run-time. We use k-Nearest Neighbors (kNN) to narrow the search space down to a subset of good candidate data points, identify the configurations that best balance the speed-quality trade-off in estimating the inverse Hessian-vector product, and introduce a fast parallel variant. Our proposed method achieves about 80X speedup while being highly correlated with the original influence values. With the availability of the fast influence functions, we demonstrate their usefulness in four applications. First, we examine whether influential data-points can "explain" test time behavior using the framework of simulatability. Second, we visualize the influence interactions between training and test data-points. Third, we show that we can correct model errors by additional fine-tuning on certain influential data-points, improving the accuracy of a trained MultiNLI model by 2.5% on the HANS dataset. Finally, we experiment with a similar setup but fine-tuning on datapoints not seen during training, improving the model accuracy by 2.8% and 1.7% on HANS and ANLI datasets respectively. Overall, our fast influence functions can be efficiently applied to large models and datasets, and our experiments demonstrate the potential of influence functions in model interpretation and correcting model errors. Code is available at https://github.com/salesforce/fast-influence-functions

arXiv comment18 pages

References in corpus (20)

• Towards A Rigorous Science of Interpretable Machine Learning
• Axiomatic Attribution for Deep Networks
• Learning Important Features Through Propagating Activation Differences
• Understanding Black-box Predictions via Influence Functions
• Explaining Machine Learning Classifiers through Diverse Counterfactual Explanations
• SmoothGrad: removing noise by adding noise
• Understanding Neural Networks through Representation Erasure
• Deep k-Nearest Neighbors: Towards Confident, Interpretable and Robust Deep Learning
• e-SNLI: Natural Language Inference with Natural Language Explanations
• Pathologies of Neural Models Make Interpretations Difficult
• Understanding the Origins of Bias in Word Embeddings
• Generative Data Augmentation for Commonsense Reasoning
• Stronger Data Poisoning Attacks Break Data Sanitization Defenses
• On the Accuracy of Influence Functions for Measuring Group Effects
• Fine-grained Sentiment Analysis with Faithful Attention
• What Would Elsa Do? Freezing Layers During Transformer Fine-Tuning
• Explaining and Improving Model Behavior with k Nearest Neighbor Representations
• Interpretable Image Recognition with Hierarchical Prototypes
• xGEMs: Generating Examplars to Explain Black-Box Models
• Pair the Dots: Jointly Examining Training History and Test Stimuli for Model Interpretability

Cited by in corpus (1)

• Post-hoc Interpretability for Neural NLP: A Survey