摘要：
## Abstract 背景： 1. the de facto standard to assess the quality of DNNs in the industry is to check their performance (accuracy) on a collected set of 阅读全文
posted @ 2023-08-29 22:25
雪溯
阅读(30)
评论(0)
推荐(0)
摘要：
## Abstract ## 1. Intro ## 2. Background ### 2.1 Program Understanding and Generation Tasks ### 2.2 NL-PL Pre-Trained Models 
评论(0)
推荐(0)
摘要：
## Abstract Task: an artifact to accompany Reducing DNN Properties to Enable Falsification with Adversarial Attacks 包括DNNF ⼯具、数据和脚本 a VirtualBox virtu 阅读全文
posted @ 2023-08-29 16:10
雪溯
阅读(19)
评论(0)
推荐(0)
摘要：
## Abstract 本文：CRADLE Task: finding and localizing bugs in DL Libraries Method: 1. differential testing between DL Libraries 2. 使用异常传播跟踪和分析来定位DL库中的错误函 阅读全文
posted @ 2023-08-29 16:10
雪溯
阅读(27)
评论(0)
推荐(0)
摘要：
## Abstract 本文：Audee Github: https://github.com/jikechao/Audee Task: test DL frameworks and localizing bugs Method: 1. search-based 2. 3 different mut 阅读全文
posted @ 2023-08-29 16:09
雪溯
阅读(34)
评论(0)
推荐(0)
摘要：
## Abstract 本文: LEMON Task: Fuzzing DL Libraries Method: 1. specified mutators for DL Libraries to explore different invoking sequences of library cod 阅读全文
posted @ 2023-08-29 16:09
雪溯
阅读(17)
评论(0)
推荐(0)
摘要：
## Abstract 本文：Myia Github: https://github.com/mila-iqia/myia Task: Review automatic differentiation for array programming in machine learning, propos 阅读全文
posted @ 2023-08-29 16:08
雪溯
阅读(17)
评论(0)
推荐(0)
摘要：
## Abstract 本文: review on adversarial attacks on CV models Task: 1. design of attack 2. defenses 3. evaluate adversarial attacks in real-world scenari 阅读全文
posted @ 2023-08-29 16:07
雪溯
阅读(18)
评论(0)
推荐(0)
摘要：
## Abstract Github: https://github.com/lts4/deepfool 背景：neural networks对小的扰动的表现并不稳定 本文: Deepfool Task: compute perturbations that fool deep networks 阅读全文
posted @ 2023-08-29 16:06
雪溯
阅读(25)
评论(0)
推荐(0)
摘要：
## Abstract 本文: DeepCheck Task: Testing DNNs with lightweight symbolic execution and program analysis Method: use lightweight symbolic analysis to ide 阅读全文
posted @ 2023-08-29 16:06
雪溯
阅读(14)
评论(0)
推荐(0)
摘要：
## Abstract 背景：大多数defenses against adversarial立刻被证明不可用 本文: review on defenses against adversarial attacks Task: 1. method 2. commonly adopted best pra 阅读全文
posted @ 2023-08-29 16:06
雪溯
阅读(23)
评论(0)
推荐(0)
摘要：
## Abstract 本文： Github: 1. https://github.com/MadryLab/mnist_challenge 2. https://github.com/MadryLab/cifar10_challenge Task: 1. study the adversarial 阅读全文
posted @ 2023-08-29 16:06
雪溯
阅读(14)
评论(0)
推荐(0)
Proj CDeepFuzz Paper Reading: Exposing numerical bugs in deep learning via gradient back-propagation
摘要：
## Abstract 背景：numerical bugs可能导致NaN或者Inf这样的异常值，这种异常值被传播后最后会造成如log()这类的函数崩溃 本文：GRIST Github: https://github.com/Jacob-yen/GRIST Task: generate a small 阅读全文
posted @ 2023-08-29 16:05
雪溯
阅读(35)
评论(0)
推荐(0)
摘要：
## Abstract 本文： Task: formalize the space of adversaries against DNNs and then introduce an adversarial testing 实验： 方法：defining a hardness measure 效果： 阅读全文
posted @ 2023-08-29 16:05
雪溯
阅读(19)
评论(0)
推荐(0)
摘要：
## Abstract 背景： the robustness requires the model to produce consistent decisions given minorly perturbed code inputs 本文:CARROT Github: https://github 阅读全文
posted @ 2023-08-29 16:05
雪溯
阅读(14)
评论(0)
推荐(0)
摘要：
## Abstract 阅读全文
posted @ 2023-08-29 16:04
雪溯
阅读(20)
评论(0)
推荐(0)
摘要：
## Abstract 本文: DeepPERF Github: https://dlperf.github.io/ 对象：TensorFlow, Keras Task: 1. characterize symptoms, root causes, and introducing and expos 阅读全文
posted @ 2023-08-29 16:04
雪溯
阅读(26)
评论(0)
推荐(0)
Proj CDeepFuzz Paper Reading: A Brief Introduction to Automatic Differentiation for Machine Learning
摘要：
## Abstract 本文：描述AD, its motivations, implementation approaches, dataflow programming, example programs with Tensorflow and PyTorch ## 1. Intro ## 2. 阅读全文
posted @ 2023-08-29 16:03
雪溯
阅读(16)
评论(0)
推荐(0)
摘要：
## Abstract 背景： 1. AD涉及computational fluid dynamics, atmospheric sciences, engineering design optimization 2. AD与DL长时间并不相互交流，直到dynamic computational g 阅读全文
posted @ 2023-08-29 16:03
雪溯
阅读(21)
评论(0)
推荐(0)
摘要：
## Abstract 背景：在architecture level detecting bugs获利更高 本文：DEBAR Github: https://github.com/ForeverZyh/DEBAR Task: static analysis of neural architectur 阅读全文
posted @ 2023-08-29 16:03
雪溯
阅读(21)
评论(0)
推荐(0)
摘要：
## Abstract 本文: Pythia Task: use static analysis to track the shapes of tensors across Python library class and warns of several possible mismatches M 阅读全文
posted @ 2023-08-29 16:03
雪溯
阅读(13)
评论(0)
推荐(0)
摘要：
## Abstract 本文：review on autoML emphasis on unsupervised anomaly detection ## 1. Intro ## 2. AutoML ### 2.1 Challenges ### 2.2 Model generation ## 3. 阅读全文
posted @ 2023-08-29 16:02
雪溯
阅读(20)
评论(0)
推荐(0)
摘要：
## Abstract 背景：由于收集到的数据，模型会学习到对性别、职业、国籍、种族的偏见，需要发现这些fairness bugs 本文：BiasFinder Task: Discover fairness bugs in the Sentiment Analysis System via Meta 阅读全文
posted @ 2023-08-29 16:02
雪溯
阅读(27)
评论(0)
推荐(0)
摘要：
## Abstract 背景：关于test RNN-based stateful system的研究较少 本文：DeepStellar Method: 1. model RNN as an abstract state transition system 2. design 2 trace simi 阅读全文
posted @ 2023-08-29 16:02
雪溯
阅读(39)
评论(0)
推荐(0)
摘要：
## Abstract 本文：VerifyML Task: testing ML based applications(ML based image classifier) by metamorphic testing Github: https://github.com/verml/VerifyM 阅读全文
posted @ 2023-08-29 16:02
雪溯
阅读(18)
评论(0)
推荐(0)
摘要：
## Abstract 本文： 1. 分析DL Library中的错误，给出了2个研究问题的答案 1. RQ1: TensorFlow 中的错误有哪些症状和原因？ 2. RQ2: TensorFlow 内部的错误在哪⾥？ 2. 总结了5个发现 1. 与症状相⽐，根本原因更具决定性，因为多个根本原因主 阅读全文
posted @ 2023-08-29 16:02
雪溯
阅读(22)
评论(0)
推荐(0)
摘要：
## Abstract 本文:RTIs(Referentially Transparent Inputs), Purity Task: Testing Machine Translation Model Method: referentially transparent input(在不同的上下文中 阅读全文
posted @ 2023-08-29 16:01
雪溯
阅读(14)
评论(0)
推荐(0)
摘要：
## Abstract Background: 1. discriminatory inputs, e.g., from societal bias, produce error, need to conduct fairness testing(generating discriminatory 阅读全文
posted @ 2023-08-29 15:48
雪溯
阅读(20)
评论(0)
推荐(0)
Proj CDeepFuzz Paper Reading: Detecting TensorFlow program bugs in real-world industrial environment
摘要：
## Abstract 本文: ShapeTracer Task: 1. Study on 12289 failed TensorFlow jobs 2. detecting TensorFlow shape-related errors with a constraint-based approa 阅读全文
posted @ 2023-08-29 15:48
雪溯
阅读(20)
评论(0)
推荐(0)
摘要：
## Abstract 本文:Metamorphic Testing for machine learning classifiers Method: cross-validation, metamorphic testing 1. MR-0: Consistence with affine tra 阅读全文
posted @ 2023-08-29 15:48
雪溯
阅读(21)
评论(0)
推荐(0)
摘要：
## Abstract 背景：Neural networks can be regarded as a new programming paradigm, Tensorflow 和PyTorch相当于Compiler，我们已知如果编译器缺乏Specification 本文：ExAIS Task: p 阅读全文
posted @ 2023-08-29 15:47
雪溯
阅读(23)
评论(0)
推荐(0)
摘要：
## Abstract 本文：Eagle Task: use equivalent graphs to differential test DL libraries Method: 使用不同的APIs, data types or optimizations来获取equivalent graphs， 阅读全文
posted @ 2023-08-29 15:47
雪溯
阅读(25)
评论(0)
推荐(0)
摘要：
## Abstract 背景：目前的研究集中在例如交换几个像素这种digital perturbation，这些是很难发生在现实世界的；生成both digital and physical adversarial perturbation很重要 本文：DeepBillboard Task: Tes 阅读全文
posted @ 2023-08-29 15:47
雪溯
阅读(28)
评论(0)
推荐(0)
摘要：
## Abstract 本文: KONURE，以及regenerator Task: use active learning to infer models of applications that retrieve data from relational databases 方法： 1. a d 阅读全文
posted @ 2023-08-29 15:46
雪溯
阅读(17)
评论(0)
推荐(0)
摘要：
## Abstract 背景： 1. Q: active learning inference based framework能够利用modularity来处理large applications中的develeopment correctness, performance and cost 2. 阅读全文
posted @ 2023-08-29 15:46
雪溯
阅读(15)
评论(0)
推荐(0)
摘要：
## Abstract 本文: RULF Github: https://github.com/Artisan-Lab/RULF Task: Library harness generation for Rust Library via API dependency graph traversal 阅读全文
posted @ 2023-08-29 15:46
雪溯
阅读(42)
评论(0)
推荐(0)
摘要：
## Abstract 背景： 1. 软件供应链攻击的目标是集成到客户端应用程序中的组件。 2. 此类攻击通常针对广泛使用的组件，通过不影响客户端观察到的组件行为（例如文件系统或网络访问）进行攻击。 本文：HARP Task: infer and regenerate the client-obse 阅读全文
posted @ 2023-08-29 15:46
雪溯
阅读(22)
评论(0)
推荐(0)
摘要：
## Abstract 背景： 1. unsafe能够绕开rust type system 2. rust libraries中常有许多unsafe keyword 本文：SyRust Task: fuzz Rust library APIs Challenge: synthesize well-t 阅读全文
posted @ 2023-08-29 15:44
雪溯
阅读(38)
评论(0)
推荐(0)
摘要：
## Abstract 背景：现有的深度学习测试在很⼤程度上依赖于⼿动标记的数据，因此通常⽆法暴露罕⻅输⼊的错误⾏为。 本文：DeepXplore Task: a white-box framework to test DL Models 方法： 1. neuron coverage 2. diff 阅读全文
posted @ 2023-08-29 12:59
雪溯
阅读(90)
评论(0)
推荐(0)
浙公网安备 33010602011771号