Deep neural networks (DNNs) are increasingly used in critical applications like autonomous vehicles and medical diagnosis, where accuracy and reliability are crucial. However, debugging DNNs is challenging and expensive, often leading to unpredictable behavior and performance issues. Identifying and diagnosing bugs in DNNs is difficult due to complex and obscure failure symptoms, which are data-driven and compute-intensive. To address this, we propose TransBug a framework that combines transformer models for feature extraction with deep learning models for classification to detect and diagnose bugs in DNNs. We employ a pre-trained transformer model, which has been trained in programming languages, to extract semantic features from both faulty and correct DNN models. We then use these extracted features in a separate deep-learning model to determine whether the code contains bugs. If a bug is detected, the model further classifies the type of bug. By leveraging the powerful feature extraction capabilities of transformers, we capture relevant characteristics from the code, which are then used by a deep learning model to identify and classify various types of bugs. This combination of transformer-based feature extraction and deep learning classification allows our method to accurately link bug symptoms to their causes, enabling developers to take targeted corrective actions. Empirical results show that the TransBug shows an accuracy of 81% for binary classification and 91% for classifying bug types.

Identifying, localizing, and resolving bugs in software engineering is challenging and costly. Approaches to resolve software bugs range from Large Language Model (LLM) code analysis and repair, and automated code repair technology that aims to alleviate the technical burden of difficult to solve bugs. We propose RAGFix, which enhances LLM’s capabilities for bug localization and code repair using Retrieval Augmented Generation (RAG) based on dynamically collected Stack Overflow posts. These posts are searchable via a Question and Answer Knowledge Graph (KGQA). We evaluate our method on the HumanEvalFix benchmark for Python using relevant closed and open-source models. Our approach facilitates error resolution in Python coding problems by creating a searchable, embedded knowledge graph representation of bug and solution information from Stack Overflow, interlinking bugs, and solutions through semi-supervised graph construction methods. We use cosine similarity on embeddings based on LLM-synthesized summaries and algorithmic features describing the coding problem and potential solution to find relevant results that improve LLM in-context performance. Our results indicate that our system enhances small open-source models’ ability to effectively repair code, particularly where these models have less parametric knowledge about relevant coding problems and can leverage nonparametric knowledge to provide accurate, actionable fixes.

The proliferation of user-friendly deepfake creation tools poses a serious challenge, demanding robust and adaptable detection strategies. Existing approaches primarily focus on raw data analysis or identifying learned artifacts or manual data-driven rules resulting in the mis-classification of deepfakes with distorted facial poses. These architectures also neglect the potential power of combining learned visual features with explicit rules.To address this gap, we introduce RuleBoost, a novel NeuroSymbolic AI based framework that seamlessly fuses extracted visual features with automatically learned rules. Our framework employs a scalable rule-based learning approach to extract learned rules from facial geometry such as distance, area, and angle. The extracted rules integrated with deep visual features show promising results giving state-of-the-art area-under-the-curve of 96.19% and 95.44% on WLDR and FaceForensics++ Datasets respectively, surpassing other deep learning specific methods. To figure out the difference NeuroSymbolic approach makes, we also analyze the samples misclassified by traditional DL-based architectures and correctly classified by Rule- Boosted architecture. Based on empirical evidence, we conclude that DL-based architectures struggle to accurately detect real and fake samples when facial artifacts lead to poses that deviate from standard facial positioning, while RuleBoost exhibits improved performance in the same scenario.