## Embracing an Adversarial Mindset for C++ Security Amanda Rousseau 2024 September 15-20 Aurora, Colorado, USA 1. Adversarial Scenarios 2. Vulnerability Trends 3. Exploits in the Wild 4. Strategies cross-org issues to combat a whole bug class 15% • Writing tools to help with discovery 4% ## Adversarial Mindset Not taught in traditional institutions Thinking Like an Adversary Challenging assumptions Note: Using libFuzzer on Windows without ASan is unsupported ## I mportance of a security-focused mindset Considering security from the start of a project Threat modeling at the design stage Know the

[Image](/uploads/documents/e/8/2/4/e824484ec02995430aad71c2a66a8be7/p39_6.jpg) ## Embracing an Adversarial Mindset for C++ Security Development Daniel Marjamäki ## KEYNOTE 14:00 15:15 Thursday 09:00 [Image](/uploads/documents/e/8/2/4/e824484ec02995430aad71c2a66a8be7/p40_1.jpg) Embracing an Adversarial Mindset for C++ Security Amanda Rousseau ![Image](/uploads/documents/e/8/2/4/e824484ec02995430aad71c2a66a8be7/p40_2

no time to review [Graham Cormode: How NOT to review a paper: the tools and techniques of the adversarial reviewer. SIGMOD Rec. 37(4) 2008]  ■ Make the paper

"sense of history." How should we choose when we have more such a sense of history and more of a mindset that we will be looked back at in the future? Furthermore, how should we act? 当然，千里之行，始于足下。那个我们 7625383934240|1726| |9|cloud-hypervisor/cloud-hypervisor|499.94850252068700|638| |10|Trusted-AI/adversarial-robustness-toolbox|495.1682899579140|1046| |11|JanusGraph/janusgraph|452.257302447994|788| |12|odpi/egeria|442 public accounts, publishing quality technical articles, and helping developers with a problem-solving mindset. Be responsive to and maintain the GitHub community, respond to developer needs, maintain code regularly

[Image](/uploads/documents/e/4/1/0/e410aece2f6936a42e1bd63213e93bec/p42_1.jpg) ## Generative Adversarial Nets Ian J. Goodfellow $ ^{*} $ , Jean Pouget-Abadie $ ^{\dagger} $ , Mehdi Mirza, Bing Xu, David involved in research projects. The most important one, in my opinion, is adversarial training (also called GAN for Generative Adversarial Networks). This is an idea that was originally proposed by Ian Goodfellow

(ILSVRC-201... LLR-pretrained ResNet-152 on ImageNet then adversarially trained for the unrestricted adversarial ## Text embedding MobileNet V2 ImageNet (ILSVRC-201... nnlm-ja-dim128-with-normalization Published trained on ImageNet (ILSVRC-2012-MobileNetV1) MobileNet V1 ImageNet (ILSVRC-2012... unsupervised-adversarial-training... Published by: DeepMind Updated: 10/24/2019 Other CIFAR-10 UAT++ adversarially

_3.jpg)  ## Adversarial AutoEncoders ■ Distribution of hidden code Encoder histogram  ## Adversarial AutoEncoders ■ Give more details after GAN ![Image](/uploads/documents/4/d/4/8/4d48c77d380b9

L. Metz 和 S. Chintala, Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks, 2015. ## 第11章 循环神经网络 人工智能的强力崛起，可能是人类历史上最好的事 情，也可能是最糟糕的事情。-史蒂芬·霍金 卷积神经网络利用数据的局部相关性 实现非常简单，通过在网络层中插入 Dropout 层即可实现网络连接的随机断开。 #### 12.3.3 Adversarial Auto-Encoder 为了能够方便地从某个已知的先验分布中 $ p(\mathbf{z}) $ 采样隐藏变量z，方便利用 $ p(\mathbf{z}) $ 来重建输入，对抗自编码器(Adversarial Auto-Encoder)利用额外的判别器网络(Discriminator，简称 Conference Track Proceedings, 2014. ## 第13章 生成对抗网络 我不能创造的事物，我就还没有完全理解它。—理查德·费曼 在生成对抗网络(Generative Adversarial Network，简称 GAN)发明之前，变分自编码器被认为是理论完备，实现简单，使用神经网络训练起来很稳定，生成的图片逼近度也较高，但是人眼还是可以很轻易地分辨出真实图片与机器生成的图片。

introduce techniques like Synthetic Minority Oversampling Technique $ ^{16} $ (SMOTE) and Generative Adversarial Network $ ^{17} $ (GAN) which can generate synthetic data for images. While SMOTE leverages statistical 1/f/8/31f883e49843cb9d0ee0ee8057c5fabe/p37_1.jpg) Figure 3-15: Architecture of a Generative Adversarial Network (GAN). It has three phases: discriminator training, generator training and the synthetic

项目提供支持。 目前，已经从 LF AI&Data 基金会毕业的项目有 Acumos Angel–ML Egeria Horovod ONNX。 正在孵化的项目有 Adlik、Adversarial Robustness Toolkit、AI Explainability 360 Toolkit、AI Fairness 360 Toolkit、Amundsen、DataPractices、DELTA、Elastic 0.04|136|1231|121|244|970|183| |9|uber/ludwig|Python|459.77|119|1786|120|304|620|265| |10|IBM/adversarial-robustness-toolbox|Python|448.39|87|884|205|309|1619|269| |11|pyro-pp/pyro|Python|414.42|87|983|93|380|1226|336|