☆ 4.6 Article

A Post-Training Framework for Improving the Performance of Deep Learning Models via Model Transformation

ACM TRANSACTIONS ON SOFTWARE ENGINEERING AND METHODOLOGY (2023)

相关参考文献

注意：仅列出部分参考文献，下载原文获取全部文献信息。

Proceedings Paper Computer Science, Software Engineering

Regression Fuzzing for Deep Learning Systems

Hanmo You et al.

Summary: In this paper, a novel regression fuzzing technique called DRFuzz is proposed to find regression faults between versions of DL systems. DRFuzz generates inputs that trigger diverse regression faults with high fidelity, and incorporates a diversity-oriented test criterion and the GAN model to enhance the diversity and fidelity of the found regression faults. Experimental results show that DRFuzz outperforms two state-of-the-art approaches in terms of the number of detected regression faults, with an average improvement of 1,177% and 539%.

2023 IEEE/ACM 45TH INTERNATIONAL CONFERENCE ON SOFTWARE ENGINEERING, ICSE (2023)

添加到收藏夹

Article Computer Science, Software Engineering

Machine Learning Testing: Survey, Landscapes and Horizons

Jie M. Zhang et al.

Summary: This paper provides a comprehensive survey of techniques for testing machine learning systems and analyzes trends and challenges in ML testing, offering promising research directions for the future.

IEEE TRANSACTIONS ON SOFTWARE ENGINEERING (2022)

添加到收藏夹

Proceedings Paper Automation & Control Systems

Learning to Construct Better Mutation Faults

Zhao Tian et al.

Summary: This study introduces a DL-based mutation technique, LEAM, which ensures the syntactic correctness of constructed mutation faults by extending a set of grammar rules specific to the mutation task, improving their representational ability. It also substantially boosts important downstream applications, such as test case prioritization and fault localization.

PROCEEDINGS OF THE 37TH IEEE/ACM INTERNATIONAL CONFERENCE ON AUTOMATED SOFTWARE ENGINEERING, ASE 2022 (2022)

添加到收藏夹

Proceedings Paper Automation & Control Systems

Toward Improving the Robustness of Deep Learning Models via Model Transformation

Yingyi Zhang et al.

Summary: This paper proposes a novel model training framework aiming at enhancing the universal robustness of DL models through model transformation and data augmentation strategy. Extensive evaluation on 9 DL models shows that the approach significantly outperforms existing adversarial training techniques.

PROCEEDINGS OF THE 37TH IEEE/ACM INTERNATIONAL CONFERENCE ON AUTOMATED SOFTWARE ENGINEERING, ASE 2022 (2022)

添加到收藏夹

Proceedings Paper Computer Science, Interdisciplinary Applications

Repairing Misclassifications in Neural Networks Using Limited Data

Patrick Henriksen et al.

Summary: We present a novel and efficient method for repairing a feed-forward neural network by addressing misclassified inputs without accessing the training set. We provide a formal characterisation for repairing the network and analyze its properties in terms of soundness and minimality. By introducing a gradient-based algorithm, localized modifications are made to the network's weights to fix misclassifications while minimally affecting accuracy on correctly classified inputs. We introduce an implementation, I-repair, which significantly reduces accuracy drops compared to other state-of-the-art repair approaches.

37TH ANNUAL ACM SYMPOSIUM ON APPLIED COMPUTING (2022)

添加到收藏夹

Article Computer Science, Hardware & Architecture

Attack-Aware Detection and Defense to Resist Adversarial Examples

Wei Jiang et al.

Summary: This article proposes an attack-aware detection and defense framework to counter adversarial attacks on security-critical artificial intelligent systems. It tests the performances of adversarial attacks, classifies them into six groups, and introduces a detector based on feature squeezing and CGR to detect these attacks effectively. Furthermore, it evaluates the defense performances of typical defense methods and provides defense recommendations for each type of adversarial attack.

IEEE TRANSACTIONS ON COMPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTEMS (2021)

添加到收藏夹

Article Computer Science, Hardware & Architecture

Supporting Deep Neural Network Safety Analysis and Retraining Through Heatmap-Based Unsupervised Learning

Hazem Fahmy et al.

Summary: The study reveals three major challenges with DNNs in safety-critical systems, including the potential risks of serious safety violations due to under-represented scenarios in the test set, the criticality of characterizing high-risk scenarios, and the lack of support for retraining DNNs when causes of violations are difficult to determine.

IEEE TRANSACTIONS ON RELIABILITY (2021)

添加到收藏夹

Article Computer Science, Artificial Intelligence

Knowledge Distillation: A Survey

Jianping Gou et al.

Summary: This paper provides a comprehensive survey of knowledge distillation, covering knowledge categories, training schemes, teacher-student architecture, distillation algorithms, performance comparison, and applications. It also briefly reviews challenges in knowledge distillation and discusses future research directions.

INTERNATIONAL JOURNAL OF COMPUTER VISION (2021)

添加到收藏夹

Proceedings Paper Computer Science, Software Engineering

Fair Preprocessing: Towards Understanding Compositional Fairness of Data Transformers in Machine Learning Pipeline

Sumon Biswas et al.

Summary: This paper discusses the fairness impact of data preprocessing stages in machine learning pipelines, identifying certain data transformers causing model unfairness and proposing a fairness composition method for choosing appropriate downstream transformers to mitigate unfairness.

PROCEEDINGS OF THE 29TH ACM JOINT MEETING ON EUROPEAN SOFTWARE ENGINEERING CONFERENCE AND SYMPOSIUM ON THE FOUNDATIONS OF SOFTWARE ENGINEERING (ESEC/FSE '21) (2021)

添加到收藏夹

Proceedings Paper Computer Science, Software Engineering

Bias in Machine Learning Software: Why? How? What to Do?

Joymallya Chakraborty et al.

Summary: This study addresses bias in software decisions and proposes a new solution. By eliminating biased labels and rebalancing internal distributions, it reduces bias while increasing performance and maintaining fairness.

PROCEEDINGS OF THE 29TH ACM JOINT MEETING ON EUROPEAN SOFTWARE ENGINEERING CONFERENCE AND SYMPOSIUM ON THE FOUNDATIONS OF SOFTWARE ENGINEERING (ESEC/FSE '21) (2021)

添加到收藏夹

Proceedings Paper Computer Science, Software Engineering

Fairea: A Model Behaviour Mutation Approach to Benchmarking Bias Mitigation Methods

Max Hort et al.

Summary: The study introduces a new approach for evaluating the effectiveness of bias mitigation methods in machine learning, revealing that some of these methods have poor effectiveness.

PROCEEDINGS OF THE 29TH ACM JOINT MEETING ON EUROPEAN SOFTWARE ENGINEERING CONFERENCE AND SYMPOSIUM ON THE FOUNDATIONS OF SOFTWARE ENGINEERING (ESEC/FSE '21) (2021)

添加到收藏夹

Proceedings Paper Computer Science, Software Engineering

A Comprehensive Study on Learning-Based PE Malware Family Classification Methods

Yixuan Ma et al.

Summary: In the field of PE malware classification, learning-based methods have gained popularity, but lack consistent comparisons between different approaches and suffer from degradation in performance due to concept drift. Existing methods face challenges of long prediction times and high memory consumption, hindering their adoption in industry usage.

PROCEEDINGS OF THE 29TH ACM JOINT MEETING ON EUROPEAN SOFTWARE ENGINEERING CONFERENCE AND SYMPOSIUM ON THE FOUNDATIONS OF SOFTWARE ENGINEERING (ESEC/FSE '21) (2021)

添加到收藏夹

Proceedings Paper Computer Science, Software Engineering

Attack as Defense: Characterizing Adversarial Examples using Robustness

Zhe Zhao et al.

Summary: Deep learning has been widely applied to real-world problems, but software based on deep learning is vulnerable to adversarial attacks. This work introduces a new characterization and defense framework to distinguish adversarial examples based on robustness evaluation. A(2)D is shown to be more effective than recent approaches in detecting and defending against adversarial attacks.

ISSTA '21: PROCEEDINGS OF THE 30TH ACM SIGSOFT INTERNATIONAL SYMPOSIUM ON SOFTWARE TESTING AND ANALYSIS (2021)

添加到收藏夹

Proceedings Paper Computer Science, Software Engineering

Reducing DNN Properties to Enable Falsification with Adversarial Attacks

David Shriver et al.

Summary: This paper explores the application of adversarial attack techniques for falsification of safety properties in Deep Neural Networks, proving that these attacks provide a powerful repertoire of scalable algorithms. It introduces a semantics-preserving reduction method to enable broad application of falsification, which proves to be cost-effective and scalable on a range of DNN correctness problems.

2021 IEEE/ACM 43RD INTERNATIONAL CONFERENCE ON SOFTWARE ENGINEERING (ICSE 2021) (2021)

添加到收藏夹

Proceedings Paper Computer Science, Software Engineering

Prioritizing Test Inputs for Deep Neural Networks via Mutation Analysis

Zan Wang et al.

Summary: The test input prioritization approach, PRIMA, based on intelligent mutation analysis and learning-to-rank, is effective in improving the efficiency of DNN testing, especially when labeling test inputs is costly. It significantly outperforms state-of-the-art approaches and has been successfully applied to practical autonomous-vehicle testing.

2021 IEEE/ACM 43RD INTERNATIONAL CONFERENCE ON SOFTWARE ENGINEERING (ICSE 2021) (2021)

添加到收藏夹

Proceedings Paper Computer Science, Software Engineering

Semi-supervised Log-based Anomaly Detection via Probabilistic Label Estimation

Lin Yang et al.

Summary: This study introduces a novel practical log-based anomaly detection approach, PLELog, which utilizes semi-supervised learning and probabilistic label estimation to enhance detection efficiency and accuracy. Evaluations on public datasets demonstrate the significant improvement of PLELog compared to other methods.

2021 IEEE/ACM 43RD INTERNATIONAL CONFERENCE ON SOFTWARE ENGINEERING (ICSE 2021) (2021)

添加到收藏夹

Proceedings Paper Computer Science, Software Engineering

A Large-scale Study on API Misuses in the Wild

Xia Li et al.

Summary: This paper conducts a large-scale empirical study on API misuses based on historical bug-fixing commits from GitHub, revealing the diversity of API misuses and extracting various frequent misuse patterns to complement existing detection tools. It also identifies important API misuse types, and validates the significance of mined patterns in guiding the design of future detection tools through user studies.

2021 14TH IEEE CONFERENCE ON SOFTWARE TESTING, VERIFICATION AND VALIDATION (ICST 2021) (2021)

添加到收藏夹

Article Computer Science, Software Engineering

Practical Accuracy Estimation for Efficient Deep Neural Network Testing

Junjie Chen et al.

ACM TRANSACTIONS ON SOFTWARE ENGINEERING AND METHODOLOGY (2020)

添加到收藏夹

Article Computer Science, Software Engineering

Guiding Dynamic Programing via Structural Probability for Accelerating Programming by Example

Ruyi Ji et al.

PROCEEDINGS OF THE ACM ON PROGRAMMING LANGUAGES-PACMPL (2020)

添加到收藏夹

Proceedings Paper Automation & Control Systems

Towards Building Robust DNN Applications: An Industrial Case Study of Evolutionary Data Augmentation

Haruki Yokoyama et al.

2020 35TH IEEE/ACM INTERNATIONAL CONFERENCE ON AUTOMATED SOFTWARE ENGINEERING (ASE 2020) (2020)

添加到收藏夹

Proceedings Paper Automation & Control Systems

Problems and Opportunities in Training Deep Learning Software Systems: An Analysis of Variance

Hung Viet Pham et al.

2020 35TH IEEE/ACM INTERNATIONAL CONFERENCE ON AUTOMATED SOFTWARE ENGINEERING (ASE 2020) (2020)

添加到收藏夹

Proceedings Paper Computer Science, Software Engineering

An Empirical Study of Common Challenges in Developing Deep Learning Applications

Tianyi Zhang et al.

2019 IEEE 30TH INTERNATIONAL SYMPOSIUM ON SOFTWARE RELIABILITY ENGINEERING (ISSRE) (2019)

添加到收藏夹

Proceedings Paper Automation & Control Systems

Inferring Program Transformations From Singular Examples via Big Code

Jiajun Jiang et al.

34TH IEEE/ACM INTERNATIONAL CONFERENCE ON AUTOMATED SOFTWARE ENGINEERING (ASE 2019) (2019)

添加到收藏夹

Proceedings Paper Computer Science, Software Engineering

A Comprehensive Study on Deep Learning Bug Characteristics

Md Johirul Islam et al.

ESEC/FSE'2019: PROCEEDINGS OF THE 2019 27TH ACM JOINT MEETING ON EUROPEAN SOFTWARE ENGINEERING CONFERENCE AND SYMPOSIUM ON THE FOUNDATIONS OF SOFTWARE ENGINEERING (2019)

添加到收藏夹

Proceedings Paper Computer Science, Software Engineering

DeepFL: Integrating Multiple Fault Diagnosis Dimensions for Deep Fault Localization

Xia Li et al.

PROCEEDINGS OF THE 28TH ACM SIGSOFT INTERNATIONAL SYMPOSIUM ON SOFTWARE TESTING AND ANALYSIS (ISSTA '19) (2019)

添加到收藏夹

Article Computer Science, Hardware & Architecture

AI Fairness 360: An extensible toolkit for detecting and mitigating algorithmic bias

R. K. E. Bellamy et al.

IBM JOURNAL OF RESEARCH AND DEVELOPMENT (2019)

添加到收藏夹

Proceedings Paper Computer Science, Software Engineering

Context-Aware Patch Generation for Better Automated Program Repair

Ming Wen et al.

PROCEEDINGS 2018 IEEE/ACM 40TH INTERNATIONAL CONFERENCE ON SOFTWARE ENGINEERING (ICSE) (2018)

添加到收藏夹

Article Computer Science, Artificial Intelligence

Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising

Kai Zhang et al.

IEEE TRANSACTIONS ON IMAGE PROCESSING (2017)

添加到收藏夹

Proceedings Paper Computer Science, Software Engineering

An Empirical Study on Real Bugs for Machine Learning Programs

Xiaobing Sun et al.

2017 24TH ASIA-PACIFIC SOFTWARE ENGINEERING CONFERENCE (APSEC 2017) (2017)

添加到收藏夹

Proceedings Paper Computer Science, Software Engineering

Precise Condition Synthesis for Program Repair

Yingfei Xiong et al.

2017 IEEE/ACM 39TH INTERNATIONAL CONFERENCE ON SOFTWARE ENGINEERING (ICSE) (2017)

添加到收藏夹

Article Computer Science, Software Engineering

Program Synthesis

Sumit Gulwani et al.

FOUNDATIONS AND TRENDS IN PROGRAMMING LANGUAGES (2017)

添加到收藏夹

Proceedings Paper Computer Science, Theory & Methods

Programming by Examples: PL Meets ML

Sumit Gulwani et al.

PROGRAMMING LANGUAGES AND SYSTEMS (APLAS 2017) (2017)

添加到收藏夹

Proceedings Paper Computer Science, Theory & Methods

TensorLayer: A Versatile Library for Efficient Deep Learning Development

Hao Dong et al.

PROCEEDINGS OF THE 2017 ACM MULTIMEDIA CONFERENCE (MM'17) (2017)

添加到收藏夹

Article Computer Science, Artificial Intelligence

Data preprocessing techniques for classification without discrimination

Faisal Kamiran et al.

KNOWLEDGE AND INFORMATION SYSTEMS (2012)

添加到收藏夹

Article Computer Science, Software Engineering

Automating String Processing in Spreadsheets Using Input-Output Examples

Sumit Gulwani

ACM SIGPLAN NOTICES (2011)

添加到收藏夹

Article Computer Science, Software Engineering

Simplifying and isolating failure-inducing input

A Zeller et al.

IEEE TRANSACTIONS ON SOFTWARE ENGINEERING (2002)

添加到收藏夹

© Peeref 2019-2024. All rights reserved.