Integrated Influence: Data Attribution with Baseline
By: Linxiao Yang, Xinyu Gu, Liang Sun
Potential Business Impact:
Shows which training data taught the AI best.
As an effective approach to quantify how training samples influence test sample, data attribution is crucial for understanding data and model and further enhance the transparency of machine learning models. We find that prevailing data attribution methods based on leave-one-out (LOO) strategy suffer from the local-based explanation, as these LOO-based methods only perturb a single training sample, and overlook the collective influence in the training set. On the other hand, the lack of baseline in many data attribution methods reduces the flexibility of the explanation, e.g., failing to provide counterfactual explanations. In this paper, we propose Integrated Influence, a novel data attribution method that incorporates a baseline approach. Our method defines a baseline dataset, follows a data degeneration process to transition the current dataset to the baseline, and accumulates the influence of each sample throughout this process. We provide a solid theoretical framework for our method, and further demonstrate that popular methods, such as influence functions, can be viewed as special cases of our approach. Experimental results show that Integrated Influence generates more reliable data attributions compared to existing methods in both data attribution task and mislablled example identification task.
Similar Papers
Revisiting Data Attribution for Influence Functions
Machine Learning (CS)
Shows which data taught a computer its answers.
Back to the Baseline: Examining Baseline Effects on Explainability Metrics
Artificial Intelligence
Makes AI explanations fairer by fixing a hidden bias.
Efficiently Verifiable Proofs of Data Attribution
Machine Learning (CS)
Lets anyone check if AI's data answers are real.