Abstract:As Deep Learning (DL) systems are widely deployed for mission-critical applications, debugging such systems becomes essential. Most existing works identify and repair suspicious neurons on the trained Deep Neural Network (DNN), which, unfortunately, might be a detour. Specifically, several existing studies have reported that many unsatisfactory behaviors are actually originated from the faults residing in DL programs. Besides, locating faulty neurons is not actionable for developers, while locating the faulty statements in DL programs can provide developers with more useful information for debugging. Though a few recent studies were proposed to pinpoint the faulty statements in DL programs or the training settings (e.g. too large learning rate), they were mainly designed based on predefined rules, leading to many false alarms or false negatives, especially when the faults are beyond their capabilities. In view of these limitations, in this paper, we proposed DeepFD, a learning-based fault diagnosis and localization framework which maps the fault localization task to a learning problem. In particular, it infers the suspicious fault types via monitoring the runtime features extracted during DNN model training and then locates the diagnosed faults in DL programs. It overcomes the limitations by identifying the root causes of faults in DL programs instead of neurons and diagnosing the faults by a learning approach instead of a set of hard-coded rules. The evaluation exhibits the potential of DeepFD. It correctly diagnoses 52% faulty DL programs, compared with around half (27%) achieved by the best state-of-the-art works. Besides, for fault localization, DeepFD also outperforms the existing works, correctly locating 42% faulty programs, which almost doubles the best result (23%) achieved by the existing works.

A Multiple-Criteria Ensemble Weight Strategy to Increase the Effectiveness of Deep Learning-based Fault Localization

Software Fault Localization Based on Multi-objective Feature Fusion and Deep Learning

Coarse-to-Fine Few-Shot Defect Recognition with Dynamic Weighting and Joint Metric

ALBFL: A Novel Neural Ranking Model for Software Fault Localization Via Combining Static and Dynamic Features

Adopting Pre-and Post-processing Weight Mechanisms to Improve Deep Learning-based Fault Localization.

DeepFD: Automated Fault Diagnosis and Localization for Deep Learning Programs

Fault Localization in Deep Learning-based Software: A System-level Approach

A General Noise-Reduction Framework for Fault Localization of Java Programs.

Multi-source fault identification based on combined deep learning

A Multi-Source Weighted Deep Transfer Network for Open-Set Fault Diagnosis of Rotary Machinery

An Innovative Method for Efficient Coverage Data Reduction in Multiple Fault Localization

Enhancing Fault Localization Through Ordered Code Analysis with LLM Agents and Self-Reflection

ABFL: an Autoencoder Based Practical Approach for Software Fault Localization.

A Study of Applying Deep Learning-Based Weighted Combinations to Improve Defect Prediction Accuracy and Effectiveness.

An Effective Data-Driven Approach for Localizing Deep Learning Faults

Federated Meta-Learning for Few-Shot Fault Diagnosis With Representation Encoding

A federated learning approach to mixed fault diagnosis in rotating machinery

A LambdaMart-Based High-Accuracy Approach for Software Automatic Fault Localization

A multi-scale collaborative fusion residual neural network-based approach for bearing fault diagnosis

A reinforcement ensemble deep transfer learning network for rolling bearing fault diagnosis with Multi-source domains