Estimating Training Data Influence by Tracing Gradient Descent
This method addresses the need for understanding and improving training data in machine learning processes, though it is incremental as it builds on gradient-based approaches.
The authors tackled the problem of estimating the influence of individual training examples on model predictions by introducing TracIn, a method that traces loss changes during training when specific examples are used, resulting in a scalable implementation applicable to any model trained with stochastic gradient descent.
We introduce a method called TracIn that computes the influence of a training example on a prediction made by the model. The idea is to trace how the loss on the test point changes during the training process whenever the training example of interest was utilized. We provide a scalable implementation of TracIn via: (a) a first-order gradient approximation to the exact computation, (b) saved checkpoints of standard training procedures, and (c) cherry-picking layers of a deep neural network. In contrast with previously proposed methods, TracIn is simple to implement; all it needs is the ability to work with gradients, checkpoints, and loss functions. The method is general. It applies to any machine learning model trained using stochastic gradient descent or a variant of it, agnostic of architecture, domain and task. We expect the method to be widely useful within processes that study and improve training data.