NNrepair: Constraint-based Repair of Neural Network Classifiers

TL;DR

NNrepair is a novel constraint-based method for repairing neural network classifiers by localizing faults and applying small parameter modifications, improving accuracy, security, and robustness without retraining.

Contribution

The paper introduces NNrepair, a new technique combining fault localization and constraint solving for targeted neural network repair at intermediate or last layers.

Findings

Improves accuracy by 45.56 percentage points on poisoned data.

Enhances robustness against adversarial attacks with a 10.40 percentage point gain.

Provides small overall accuracy improvements without retraining.

Abstract

We present NNrepair, a constraint-based technique for repairing neural network classifiers. The technique aims to fix the logic of the network at an intermediate layer or at the last layer. NNrepair first uses fault localization to find potentially faulty network parameters (such as the weights) and then performs repair using constraint solving to apply small modifications to the parameters to remedy the defects. We present novel strategies to enable precise yet efficient repair such as inferring correctness specifications to act as oracles for intermediate layer repair, and generation of experts for each class. We demonstrate the technique in the context of three different scenarios: (1) Improving the overall accuracy of a model, (2) Fixing security vulnerabilities caused by poisoning of training data and (3) Improving the robustness of the network against adversarial attacks. Our evaluation on MNIST and CIFAR-10 models shows that NNrepair can improve the accuracy by 45.56 percentage points on poisoned data and 10.40 percentage points on adversarial data. NNrepair also provides small improvement in the overall accuracy of models, without requiring new data or re-training.