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

Over the past decade, Deep Learning (DL) has become an integral part of our daily lives. This surge in DL usage has heightened the need for developing reliable DL software systems. Given that fault localization is a critical task in reliability assessment, researchers have proposed several fault localization techniques for DL-based software, primarily focusing on faults within the DL model. While the DL model is central to DL components, there are other elements that significantly impact the performance of DL components. As a result, fault localization methods that concentrate solely on the DL model overlook a large portion of the system. To address this, we introduce FL4Deep, a system-level fault localization approach considering the entire DL development pipeline to effectively localize faults across the DL-based systems. In an evaluation using 100 faulty DL scripts, FL4Deep outperformed four previous approaches in terms of accuracy for three out of six DL-related faults, including issues related to data (84%), mismatched libraries between training and deployment (100%), and loss function (69%). Additionally, FL4Deep demonstrated superior precision and recall in fault localization for five categories of faults including three mentioned fault types in terms of accuracy, plus insufficient training iteration and activation function.

Towards Optimizing SQL Generation via LLM Routing

Text-to-SQL enables users to interact with databases through natural language, simplifying access to structured data. Although highly capable large language models (LLMs) achieve strong accuracy for complex queries, they incur unnecessary latency and dollar cost for simpler ones. In this paper, we introduce the first LLM routing approach for Text-to-SQL, which dynamically selects the most cost-effective LLM capable of generating accurate SQL for each query. We present two routing strategies (score- and classification-based) that achieve accuracy comparable to the most capable LLM while reducing costs. We design the routers for ease of training and efficient inference. In our experiments, we highlight a practical and explainable accuracy-cost trade-off on the BIRD dataset.

In-Simulation Testing of Deep Learning Vision Models in Autonomous Robotic Manipulators

Testing autonomous robotic manipulators is challenging due to the complex software interactions between vision and control components. A crucial element of modern robotic manipulators is the deep learning based object detection model. The creation and assessment of this model requires real world data, which can be hard to label and collect, especially when the hardware setup is not available. The current techniques primarily focus on using synthetic data to train deep neural networks (DDNs) and identifying failures through offline or online simulation-based testing. However, the process of exploiting the identified failures to uncover design flaws early on, and leveraging the optimized DNN within the simulation to accelerate the engineering of the DNN for real-world tasks remains unclear. To address these challenges, we propose the MARTENS (Manipulator Robot Testing and Enhancement in Simulation) framework, which integrates a photorealistic NVIDIA Isaac Sim simulator with evolutionary search to identify critical scenarios aiming at improving the deep learning vision model and uncovering system design flaws. Evaluation of two industrial case studies demonstrated that MARTENS effectively reveals robotic manipulator system failures, detecting 25 % to 50 % more failures with greater diversity compared to random test generation. The model trained and repaired using the MARTENS approach achieved mean average precision (mAP) scores of 0.91 and 0.82 on real-world images with no prior retraining. Further fine-tuning on real-world images for a few epochs (less than 10) increased the mAP to 0.95 and 0.89 for the first and second use cases, respectively. In contrast, a model trained solely on real-world data achieved mAPs of 0.8 and 0.75 for use case 1 and use case 2 after more than 25 epochs.

Toward Debugging Deep Reinforcement Learning Programs with RLExplorer

Deep reinforcement learning (DRL) has shown success in diverse domains such as robotics, computer games, and recommendation systems. However, like any other software system, DRL-based software systems are susceptible to faults that pose unique challenges for debugging and diagnosing. These faults often result in unexpected behavior without explicit failures and error messages, making debugging difficult and time-consuming. Therefore, automating the monitoring and diagnosis of DRL systems is crucial to alleviate the burden on developers. In this paper, we propose RLExplorer, the first fault diagnosis approach for DRL-based software systems. RLExplorer automatically monitors training traces and runs diagnosis routines based on properties of the DRL learning dynamics to detect the occurrence of DRL-specific faults. It then logs the results of these diagnoses as warnings that cover theoretical concepts, recommended practices, and potential solutions to the identified faults. We conducted two sets of evaluations to assess RLExplorer. Our first evaluation of faulty DRL samples from Stack Overflow revealed that our approach can effectively diagnose real faults in 83% of the cases. Our second evaluation of RLExplorer with 15 DRL experts/developers showed that (1) RLExplorer could identify 3.6 times more defects than manual debugging and (2) RLExplorer is easily integrated into DRL applications.

What Information Contributes to Log-based Anomaly Detection? Insights from a Configurable Transformer-Based Approach

Log data are generated from logging statements in the source code, providing insights into the execution processes of software applications and systems. State-of-the-art log-based anomaly detection approaches typically leverage deep learning models to capture the semantic or sequential information in the log data and detect anomalous runtime behaviors. However, the impacts of these different types of information are not clear. In addition, existing approaches have not captured the timestamps in the log data, which can potentially provide more fine-grained temporal information than sequential information. In this work, we propose a configurable transformer-based anomaly detection model that can capture the semantic, sequential, and temporal information in the log data and allows us to configure the different types of information as the model's features. Additionally, we train and evaluate the proposed model using log sequences of different lengths, thus overcoming the constraint of existing methods that rely on fixed-length or time-windowed log sequences as inputs. With the proposed model, we conduct a series of experiments with different combinations of input features to evaluate the roles of different types of information in anomaly detection. When presented with log sequences of varying lengths, the model can attain competitive and consistently stable performance compared to the baselines. The results indicate that the event occurrence information plays a key role in identifying anomalies, while the impact of the sequential and temporal information is not significant for anomaly detection in the studied public datasets. On the other hand, the findings also reveal the simplicity of the studied public datasets and highlight the importance of constructing new datasets that contain different types of anomalies to better evaluate the performance of anomaly detection models.

Trimming the Risk: Towards Reliable Continuous Training for Deep Learning Inspection Systems

The industry increasingly relies on deep learning (DL) technology for manufacturing inspections, which are challenging to automate with rule-based machine vision algorithms. DL-powered inspection systems derive defect patterns from labeled images, combining human-like agility with the consistency of a computerized system. However, finite labeled datasets often fail to encompass all natural variations necessitating Continuous Training (CT) to regularly adjust their models with recent data. Effective CT requires fresh labeled samples from the original distribution; otherwise, selfgenerated labels can lead to silent performance degradation. To mitigate this risk, we develop a robust CT-based maintenance approach that updates DL models using reliable data selections through a two-stage filtering process. The initial stage filters out low-confidence predictions, as the model inherently discredits them. The second stage uses variational auto-encoders and histograms to generate image embeddings that capture latent and pixel characteristics, then rejects the inputs of substantially shifted embeddings as drifted data with erroneous overconfidence. Then, a fine-tuning of the original DL model is executed on the filtered inputs while validating on a mixture of recent production and original datasets. This strategy mitigates catastrophic forgetting and ensures the model adapts effectively to new operational conditions. Evaluations on industrial inspection systems for popsicle stick prints and glass bottles using critical real-world datasets showed less than 9% of erroneous self-labeled data are retained after filtering and used for fine-tuning, improving model performance on production data by up to 14% without compromising its results on original validation data.

Assessing Programming Task Difficulty for Efficient Evaluation of Large Language Models

Large Language Models (LLMs) show promising potential in Software Engineering, especially for code-related tasks like code completion and code generation. LLMs' evaluation is generally centred around general metrics computed over benchmarks. While painting a macroscopic view of the benchmarks and of the LLMs' capacity, it is unclear how each programming task in these benchmarks assesses the capabilities of the LLMs. In particular, the difficulty level of the tasks in the benchmarks is not reflected in the score used to report the performance of the model. Yet, a model achieving a 90% score on a benchmark of predominantly easy tasks is likely less capable than a model achieving a 90% score on a benchmark containing predominantly difficult tasks. This paper devises a framework, HardEval, for assessing task difficulty for LLMs and crafting new tasks based on identified hard tasks. The framework uses a diverse array of prompts for a single task across multiple LLMs to obtain a difficulty score for each task of a benchmark. Using two code generation benchmarks, HumanEval+ and ClassEval, we show that HardEval can reliably identify the hard tasks within those benchmarks, highlighting that only 21% of HumanEval+ and 27% of ClassEval tasks are hard for LLMs. Through our analysis of task difficulty, we also characterize 6 practical hard task topics which we used to generate new hard tasks. Orthogonal to current benchmarking evaluation efforts, HardEval can assist researchers and practitioners in fostering better assessments of LLMs. The difficulty score can be used to identify hard tasks within existing benchmarks. This, in turn, can be leveraged to generate more hard tasks centred around specific topics either for evaluation or improvement of LLMs. HardEval generalistic approach can be applied to other domains such as code completion or Q/A.

DeepCodeProbe: Towards Understanding What Models Trained on Code Learn

Machine learning models trained on code and related artifacts offer valuable support for software maintenance but suffer from interpretability issues due to their complex internal variables. These concerns are particularly significant in safety-critical applications where the models' decision-making processes must be reliable. The specific features and representations learned by these models remain unclear, adding to the hesitancy in adopting them widely. To address these challenges, we introduce DeepCodeProbe, a probing approach that examines the syntax and representation learning abilities of ML models designed for software maintenance tasks. Our study applies DeepCodeProbe to state-of-the-art models for code clone detection, code summarization, and comment generation. Findings reveal that while small models capture abstract syntactic representations, their ability to fully grasp programming language syntax is limited. Increasing model capacity improves syntax learning but introduces trade-offs such as increased training time and overfitting. DeepCodeProbe also identifies specific code patterns the models learn from their training data. Additionally, we provide best practices for training models on code to enhance performance and interpretability, supported by an open-source replication package for broader application of DeepCodeProbe in interpreting other code-related models.

Chain of Targeted Verification Questions to Improve the Reliability of Code Generated by LLMs

LLM-based assistants, such as GitHub Copilot and ChatGPT, have the potential to generate code that fulfills a programming task described in a natural language description, referred to as a prompt. The widespread accessibility of these assistants enables users with diverse backgrounds to generate code and integrate it into software projects. However, studies show that code generated by LLMs is prone to bugs and may miss various corner cases in task specifications. Presenting such buggy code to users can impact their reliability and trust in LLM-based assistants. Moreover, significant efforts are required by the user to detect and repair any bug present in the code, especially if no test cases are available. In this study, we propose a self-refinement method aimed at improving the reliability of code generated by LLMs by minimizing the number of bugs before execution, without human intervention, and in the absence of test cases. Our approach is based on targeted Verification Questions (VQs) to identify potential bugs within the initial code. These VQs target various nodes within the Abstract Syntax Tree (AST) of the initial code, which have the potential to trigger specific types of bug patterns commonly found in LLM-generated code. Finally, our method attempts to repair these potential bugs by re-prompting the LLM with the targeted VQs and the initial code. Our evaluation, based on programming tasks in the CoderEval dataset, demonstrates that our proposed method outperforms state-of-the-art methods by decreasing the number of targeted errors in the code between 21% to 62% and improving the number of executable code instances to 13%.

Mining Action Rules for Defect Reduction Planning

Defect reduction planning plays a vital role in enhancing software quality and minimizing software maintenance costs. By training a black box machine learning model and "explaining" its predictions, explainable AI for software engineering aims to identify the code characteristics that impact maintenance risks. However, post-hoc explanations do not always faithfully reflect what the original model computes. In this paper, we introduce CounterACT, a Counterfactual ACTion rule mining approach that can generate defect reduction plans without black-box models. By leveraging action rules, CounterACT provides a course of action that can be considered as a counterfactual explanation for the class (e.g., buggy or not buggy) assigned to a piece of code. We compare the effectiveness of CounterACT with the original action rule mining algorithm and six established defect reduction approaches on 9 software projects. Our evaluation is based on (a) overlap scores between proposed code changes and actual developer modifications; (b) improvement scores in future releases; and (c) the precision, recall, and F1-score of the plans. Our results show that, compared to competing approaches, CounterACT's explainable plans achieve higher overlap scores at the release level (median 95%) and commit level (median 85.97%), and they offer better trade-off between precision and recall (median F1-score 88.12%). Finally, we venture beyond planning and explore leveraging Large Language models (LLM) for generating code edits from our generated plans. Our results show that suggested LLM code edits supported by our plans are actionable and are more likely to pass relevant test cases than vanilla LLM code recommendations.