skip to main content

Title: Repairing Deep Neural Networks: Fix Patterns and Challenges
Significant interest in applying Deep Neural Network (DNN) has fueled the need to support engineering of software that uses DNNs. Repairing software that uses DNNs is one such unmistakable SE need where automated tools could be very helpful; however, we do not fully understand challenges to repairing and patterns that are utilized when manually repairing them. What challenges should automated repair tools address? What are the repair patterns whose automation could help developers? Which repair patterns should be assigned a higher priority for automation? This work presents a comprehensive study of bug fix patterns to address these questions. We have studied 415 repairs from Stack Overflow and 555 repairs from GitHub for five popular deep learning libraries Caffe, Keras, Tensorflow, Theano, and Torch to understand challenges in repairs and bug repair patterns. Our key findings reveal that DNN bug fix patterns are distinctive compared to traditional bug fix patterns; the most common bug fix patterns are fixing data dimension and neural network connectivity; DNN bug fixes have the potential to introduce adversarial vulnerabilities; DNN bug fixes frequently introduce new bugs; and DNN bug localization, reuse of trained model, and coping with frequent releases are major challenges faced by developers when more » fixing bugs. We also contribute a benchmark of 667 DNN (bug, repair) instances. « less
Authors:
; ; ;
Award ID(s):
1934884 1513263
Publication Date:
NSF-PAR ID:
10178780
Journal Name:
ICSE'20: The 42nd International Conference on Software Engineering
Sponsoring Org:
National Science Foundation
More Like this
  1. Static analysis tools have demonstrated effectiveness at finding bugs in real world code. Such tools are increasingly widely adopted to improve software quality in practice. Automated Program Repair (APR) has the potential to further cut down on the cost of improving software quality. However, there is a disconnect between these effective bug-finding tools and APR. Recent advances in APR rely on test cases, making them inapplicable to newly discovered bugs or bugs difficult to test for deterministically (like memory leaks). Additionally, the quality of patches generated to satisfy a test suite is a key challenge. We address these challenges by adapting advances in practical static analysis and verification techniques to enable a new technique that finds and then accurately fixes real bugs without test cases. We present a new automated program repair technique using Separation Logic. At a high-level, our technique reasons over semantic effects of existing program fragments to fix faults related to general pointer safety properties: resource leaks, memory leaks, and null dereferences. The procedure automatically translates identified fragments into source-level patches, and verifies patch correctness with respect to reported faults. In this work we conduct the largest study of automatically fixing undiscovered bugs in real-world code tomore »date. We demonstrate our approach by correctly fixing 55 bugs, including 11 previously undiscovered bugs, in 11 real-world projects.« less
  2. Deep Neural Networks (DNNs) are becoming an integral part of most software systems. Previous work has shown that DNNs have bugs. Unfortunately, existing debugging techniques don't support localizing DNN bugs because of the lack of understanding of model behaviors. The entire DNN model appears as a black box. To address these problems, we propose an approach and a tool that automatically determines whether the model is buggy or not, and identifies the root causes for DNN errors. Our key insight is that historic trends in values propagated between layers can be analyzed to identify faults, and also localize faults. To that end, we first enable dynamic analysis of deep learning applications: by converting it into an imperative representation and alternatively using a callback mechanism. Both mechanisms allows us to insert probes that enable dynamic analysis over the traces produced by the DNN while it is being trained on the training data. We then conduct dynamic analysis over the traces to identify the faulty layer or hyperparameter that causes the error. We propose an algorithm for identifying root causes by capturing any numerical error and monitoring the model during training and finding the relevance of every layer/parameter on the DNN outcome.more »We have collected a benchmark containing 40 buggy models and patches that contain real errors in deep learning applications from Stack Overflow and GitHub. Our benchmark can be used to evaluate automated debugging tools and repair techniques. We have evaluated our approach using this DNN bug-and-patch benchmark, and the results showed that our approach is much more effective than the existing debugging approach used in the state-of-the-practice Keras library. For 34/40 cases, our approach was able to detect faults whereas the best debugging approach provided by Keras detected 32/40 faults. Our approach was able to localize 21/40 bugs whereas Keras did not localize any faults.« less
  3. Effective symbolic evaluation is key to building scalable ver- ification and synthesis tools based on SMT solving. These tools use sym- bolic evaluators to reduce the semantics of all paths through a finite program to logical constraints, discharged with an SMT solver. Using an evaluator effectively requires tool developers to be able to identify and re- pair performance bottlenecks in code under all-path evaluation, a difficult task, even for experts. This paper presents a new method for repairing such bottlenecks automatically. The key idea is to formulate the symbolic performance repair problem as combinatorial search through a space of semantics-preserving transformations, or repairs, to find an equivalent program with minimal cost under symbolic evaluation. The key to real- izing this idea is (1) defining a small set of generic repairs that can be combined to fix common bottlenecks, and (2) searching for combinations of these repairs to find good solutions quickly and best ones eventually. Our technique, SymFix, contributes repairs based on deforestation and symbolic reflection, and an efficient algorithm that uses symbolic profil- ing to guide the search for fixes. To evaluate SymFix, we implement it for the Rosette solver-aided language and symbolic evaluator. Applying SymFix to 18 publishedmore »verification and synthesis tools built in Rosette, we find that it automatically improves the performance of 12 tools by a factor of 1.1×–91.7×, and 4 of these fixes match or outperform expert- written repairs. SymFix also finds 5 fixes that were missed by experts.« less
  4. Of the hundreds of billions of dollars spent on developer wages, up to 25% accounts for fixing bugs. Companies like Google save significant human effort and engineering costs with automatic bug detection tools, yet automatically fixing them is still a nascent endeavour. Very recent work (including our own) demonstrates the feasibility of automatic program repair in practice. As automated repair technology matures, it presents great appeal for integration into developer workflows. We believe software bots are a promising vehicle for realizing this integration, as they bridge the gap between human software development and automated processes. We envision repair bots orchestrating automated refactoring and bug fixing. To this end, we explore what building a repair bot entails. We draw on our understanding of patch generation, validation, and real world software development interactions to identify six principles that bear on engineering repair bots and discuss related design challenges for integrating human workflows. Ultimately, this work aims to foster critical focus and interest for making repair bots a reality.
  5. Deep learning has gained substantial popularity in recent years. Developers mainly rely on libraries and tools to add deep learning capabilities to their software. What kinds of bugs are frequently found in such software? What are the root causes of such bugs? What impacts do such bugs have? Which stages of deep learning pipeline are more bug prone? Are there any antipatterns? Understanding such characteristics of bugs in deep learning software has the potential to foster the development of better deep learning platforms, debugging mechanisms, development practices, and encourage the development of analysis and verification frameworks. Therefore, we study 2716 high-quality posts from Stack Overflow and 500 bug fix commits from Github about five popular deep learning libraries Caffe, Keras, Tensorflow, Theano, and Torch to understand the types of bugs, root causes of bugs, impacts of bugs, bug-prone stage of deep learning pipeline as well as whether there are some common antipatterns found in this buggy software. The key findings of our study include: data bug and logic bug are the most severe bug types in deep learning software appearing more than 48% of the times, major root causes of these bugs are Incorrect Model Parameter (IPS) and Structural Inefficiencymore »(SI) showing up more than 43% of the times.We have also found that the bugs in the usage of deep learning libraries have some common antipatterns.« less