论文阅读记录

jshiro 发布于 2024-10-24 收录于记录

论文	开源	阅读状态	是否复现	类型
LLM-Fuzzer: Scaling Assessment of Large Language Model Jailbreaks	https://github.com/sherdencooper/GPTFuzz	sc	n	大模型模糊测试
FairECom: Towards Proof of E-Commerce Fairness Against Price Discrimination	https://github.com/CyberSec-Dev/FairECom	pc	n	价格歧视规避
PyRTFuzz: Detecting Bugs in Python Runtimes via Two-Level Collaborative Fuzzing	https://github.com/awen-li/PyRTFuzz	n	n	Python模糊测试
Towards Robust Detection of Open Source Software Supply Chain Poisoning Attacks in Industry Environments	-	fc	n	软件供应链投毒攻击检测
Cognitive Bias in Decision-Making with LLMs	https://huggingface.co/datasets/jecht/cognitive_bias 数据集	fc	n	大模型认知偏差检测
Generative AI for Game Theory-based Mobile Networking	-	fc	n	博弈论，生成式AI

阅读状态：fc(Fully completed), sc(Substantially completed), pc(Partially completed)

计算机视觉

jshiro 发布于 2024-10-12 收录于 Notes AI

N x N * f x f = ( N - f + 1 ) * ( N - f + 1)，f 基本为奇数

滤波器/卷积核

[-1, 0, 1]
[-2, 0, 2]
[-1, 0, 1]

[-3, 0, 3]
[-10, 0, 10]
[-3, 0, 3]

padding填充

jshiro 发布于 2024-10-12 收录于 Notes AI

要素：$state\overset{policy}\longrightarrow action$

奖励函数

jshiro 发布于 2024-10-12 收录于 Notes AI

整体框架

特点

参数随机初始化

jshiro 发布于 2024-10-12 收录于 Notes AI

自然语言处理，Natural language processing (NLP) ，有监督学习

文本是一种序列数据，一句话是由一个个按顺序排列的单词组成的

数据表示

$X^{(i)} = [x_1^{(i)}, x_2^{(i)}, \dots, x_{T_x^{(i)}}^{(i)}]$：第 $i$ 个样本的输入序列，有 $T_x^{(i)}$ 个元素（单词/字符等）

jshiro 发布于 2024-10-09 收录于 Notes

bash的配置在~/.bashrc中

git的配置在~/.gitconfig中

vim的配置在~/.vimrc中

tree：列举所有文件，递归列举文件