526互联

Paper

Proj CDeepFuzz Paper Reading: Balancing Effectiveness and Flakiness of Non-Deterministic Machine Learning Tests

## Abstract 背景：In fact, some of the latest findings suggest that the existence of adversarial attacks may be an inherent weakness of deep learning mod ......
Non-Deterministic Effectiveness Deterministic CDeepFuzz Balancing

Proj CDeepFuzz Paper Reading: NeuRI: Diversifying DNN Generation via Inductive Rule Inference

## Abstract 背景：The correctness of DL systems is crucial for trust in DL applications 本文: NeuRI BaseTool: FreeFuzz Github: https://github.com/ise-uiuc/ ......
Diversifying Generation CDeepFuzz Inductive Inference

Proj CDeepFuzz Paper Reading: Natural attack for pre-trained models of code

## Abstract 背景：目前大多数的adversarial attack method on pre-trained models of code忽略了perturbations should be natural to human judges(naturalness requirement ......
pre-trained CDeepFuzz Natural Reading trained

Proj CDeepFuzz Paper Reading: COMET: Coverage-guided Model Generation For Deep Learning Library Testing

## Abstract 背景：已有的方法(Muffin, Lemon, Cradle) can cover at most 34.1% layer inputs, 25.9% layer parameter values, and 15.6% layer sequences. 本文：COMET Gi ......
Coverage-guided Generation CDeepFuzz Coverage Learning

Proj CDeepFuzz Paper Reading: IvySyn: Automated Vulnerability Discovery in Deep Learning Frameworks

## Abstract 本文：IvySyn Task: discover memory error vulnerabilities in DL frameworks BugType: memory safety errors, fatal runtime errors Method: 1. 利用na ......
Vulnerability Frameworks CDeepFuzz Automated Discovery

MAPF Paper Reading Note

随便写写记录一下 ## 1. 2005-Cooperative Pathfinding ### 1.1. LRA* local repair A* - 依次做A* - 即将开始碰撞时，replan - a general replan solution: 每次重规划时，新增noise，按照比例加入$ ......
Reading Paper MAPF Note

Proj CDeepFuzz Paper Reading: Invariance-inducing regularization using worst-case transformations suffices to boost accuracy and spatial robustness

## Abstract 本文： Task: 1. prove invariance-inducing regularizers can increase predictive accuracy for worst-case spatial transformations 2. prove that ......
Invariance-inducing transformations regularization Invariance robustness

Proj CDeepFuzz Paper Reading: Framework for Evaluating Faithfulness of Local Explanations

## Abstract 本文: Task: 1. study the faithfulness of an explanation system to the underlying prediction model on consistency and sufficiency 2. introduc ......
Faithfulness Explanations Evaluating CDeepFuzz Framework

Proj CDeepFuzz Paper Reading: Differential Testing of Cross Deep Learning Framework APIs: Revealing Inconsistencies and Vulnerabilities

## Abstract 背景：目前对cross-framework conversion中的inconsistencies和security bugs的研究少有 本文：TensorScope Task: test cross-frame APIs in Machine Learning Librar ......
Inconsistencies Vulnerabilities Differential CDeepFuzz Framework

Proj CDeepFuzz Paper Reading: DeepTest: automated testing of deep-neural-network-driven autonomous cars

## Abstract 本文: DeepTest Task: a systematic testing tool for DNN-driven vehicles Method: 1. generated test cases with real-world changes like rain, fo ......
deep-neural-network-driven autonomous CDeepFuzz automated DeepTest

Proj CDeepFuzz Paper Reading: DeepGauge: multi-granularity testing criteria for deep learning systems

## Abstract 本文: DeepGauge Task: provide multi-granularity testing criteria for DL systems Method: multi-granularity testing criteria for DL systems: 1 ......
multi-granularity granularity CDeepFuzz DeepGauge criteria

Proj CDeepFuzz Paper Reading: Combinatorial Testing for Deep Learning Systems

## Abstract 本文：DeepCT Task: Testing DL Models with Combinatorial Testing Method: 1. 将输出值的空间离散化为区间，以便覆盖每个区间,对不同层内的神经元交互进⾏采样，并减少必须执⾏的测试输⼊的数量。 2. a set o ......
Combinatorial CDeepFuzz Learning Reading Testing

Proj CDeepFuzz Paper Reading: Automatic differentiation in PyTorch

## Abstract 本文：描述automatic differentiation module of PyTorch 包括：Lua Torch, Chainer, HIPS Autograd Task: Provides a high-performance environment on dif ......
differentiation CDeepFuzz Automatic Reading PyTorch

Proj CDeepFuzz Paper Reading: ACETest: Automated Constraint Extraction for Testing Deep Learning Operators

## Abstract Github: https://github.com/shijy16/ACETest 背景： 1. DL operators 用来计算多维tensors，很重要 本文：ACETest Task: automatically extract input validation c ......
Constraint Extraction CDeepFuzz Automated Operators

Proj CDeepFuzz Paper Reading: AutoML: A survey of the state-of-the-art

## Abstract Github: https://github.com/marsggbo/automl_a_survey_of_state_of_the_art 本文： 1. intro AutoML methods: data preparation, feature engineering ......
state-of-the-art CDeepFuzz the Reading AutoML

Proj CDeepFuzz Paper Reading: Aries: Efficient Testing of Deep Neural Networks via Labeling-Free Accuracy Estimation

## 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 ......
Labeling-Free Estimation CDeepFuzz Efficient Accuracy

Proj CDeepFuzz Paper Reading: An Extensive Study on Pre-trained Models for Program Understanding and Generation

## Abstract ## 1. Intro ## 2. Background ### 2.1 Program Understanding and Generation Tasks ### 2.2 NL-PL Pre-Trained Models ![](https://img2023.cnblo ......
Understanding Pre-trained Generation CDeepFuzz Extensive

Proj CDeepFuzz Paper Reading: SyRust: automatic testing of Rust libraries with semantic-aware program synthesis

## 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 ......
semantic-aware CDeepFuzz automatic libraries synthesis

Proj CDeepFuzz Paper Reading: Deepxplore: Automated whitebox testing of deep learning systems

## 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 ......
Deepxplore CDeepFuzz Automated whitebox learning

Paper Reading: PCTBagging: From inner ensembles to ensembles. A trade-off between discriminating capacity and interpretability

针对 CTC 分类性能较差和 Bagging 的可解释性较差的问题，本文提出了一种结合 CTC 和 Bagging 的算法 PCTBagging。首先构建一棵不完整的 CTC，CTC 的规模由超参数合并比来确定，接着使用 Bagging 完成后续的树结构的生成。将 PCTBagging 的结果与 B... ......
ensembles interpretability discriminating PCTBagging trade-off

Paper Reading：ControlBurn-Feature Selection by Sparse Forests

针对存在大量相关特征时重要特征的影响被削弱的问题，本文设计了一种通过稀疏森林来消除相关偏差的特征选择算法 ControlBurn。首先使用套袋和提升等方法生成森林，然后通过一个平衡特征稀疏性和预测性能的组 LASSO 惩罚目标为每棵树选择稀疏权值，从而减少树的数量。与 Wrapper 特征选择方法不... ......
ControlBurn-Feature ControlBurn Selection Feature Forests

Paper Reading 博客汇总

按照算法的类型对个人的 Paper Reading 博客进行汇总，涉及多个研究方向的论文将按照个人主观感觉的主要方向排列。 [toc] #不平衡学习 * [A Novel Model for Imbalanced Data Classification](https://www.cnblogs.co ......
Reading Paper 博客

Paper Reading: A pareto-based ensemble of feature selection algorithms

本文将集成特征选择问题建模为具有两个目标的帕累托优化问题，提出一种类型的异构集成特征选择算法 PEFS。首先采用两种聚合方法对四种不同 FS 方法得到的结果进行组合，接着使用双目标优化来评估这些结果，最后根据非优势特征在双目标空间中的拥挤距离进行排序。该方法平衡了关联度和冗余性两种不同的 FS 方法... ......
pareto-based algorithms selection ensemble Reading

Paper Reading: NBDT: Neural-Backed Decision Trees

为了提高计算机视觉模型的可解释性，本文融合深度学习和决策树提出了神经支持决策树(NBDTs)。NBDT 使用一个可微的倾斜决策树取代了神经网络的最后一个线性层，和经典的决策树方法不同，NBDT 使用从模型参数派生的层次结构，不使用分层 softmax。NBDT 可以从任何现有的分类神经网络中创建，无... ......
Neural-Backed Decision Reading Backed Neural

Paper Reading: Multitree Genetic Programming With New Operators for Transfer Learning in Symbolic Regression With Incomplete Data

针对数据集存在缺失值的问题，本文提出了一种基于多树 GP(MTGP) 的迁移学习方法 pMTGPDA，用于将知识从完整的源域转移到不完整的目标域中。首先在源域的数据集上训练多个 SR 模型，通过模型中的训练细节计算源域的特征和实例的权重作为先验知识。然后将提取的权重知识用于基于 MTGP 的转换，构... ......
With Programming Incomplete Regression Multitree

Paper Reading: FT4cip: A new functional tree for classification in class imbalance problems

本文提出了一种类不平衡问题的功能树(FT4cip)，该模型使用了考虑类不平衡的分割评估函数 Twoing，以及使用了一种优化 AUC 的新型剪枝算法。同时对多变量分割使用特征选择，进一步提高分类性能和可解释性。通过大量的实验分析证明，FT4cip 在 AUC 上的分类性能优于 LMT 和 Gama。... ......
classification functional imbalance problems Reading

【JointJS】Paper 事件和 Graph 事件

# 复习 Paper & Graph Paper 是渲染我们提供的数据的，表示的是视图层（View）。数据和逻辑在 Graph 中操作，表示的是数据层和控制层（Model & Controller）。 ![Graph & Paper 示意图](https://img2023.cnblogs.com/ ......
事件 JointJS Graph Paper

Paper Reading: A Re-Balancing Strategy for Class-Imbalanced Classification Based on Instance Difficulty

受人类学习过程的启发，本文根据学习速度设计了样本难度模型，并提出了一种新的实例级再平衡策略。具体来说模型在每个训练周期记录每个实例的预测，并根据预测的变化来测量该样本的难度难度。然后对困难实例赋予更高的权重，对数据进行重新采样。本文从理论上证明了提出的重采样策略的正确性和收敛性，并进行一些实证实验来... ......
Class-Imbalanced Classification Re-Balancing Imbalanced Difficulty

Paper Reading: Exploratory Undersampling for Class-Imbalance Learning

本文是不平衡分类问题的经典论文，文中提出了 2 种不平衡集成学习模型都是简单而有效的 baseline 方法。 EasyEnsemble 方法直接对多数类样本进行采样得到几个子集，并使用这些子集分别训练基分类器。BalanceCascade 是使用训练好的分类器来指导后续分类器的采样过程，即在上一个... ......
Class-Imbalance Undersampling Exploratory Imbalance Learning

Proj. CMI Paper Reading: Conversational Automated Program Repair

## Abstract 背景：之前APR与大模型的结合仅仅使用了一些特殊构建的input/prompt 本文：conversational APR 方法：使用long-term context window来记录之前的validation feedback 实验： 在10个LLMs上进行对比展示改进 ......
Conversational Automated Program Reading Repair
共82篇  :2/3页 首页上一页2下一页尾页