CodePath:算法设计与分析

宋圣洁(Mr. Holy Sung)

HomePage

个人主页:宋圣洁

Email

sjs@mail.bnu.edu.cn
shengjiesong6@gmail.com

Education

Peking University, Master of Software Engineering (2025 – 2028)

Beijing Normal University, Bachelor of Finance Technology (2021 – 2025)
• GPA: 3.9/4.0

The Hong Kong University of Science and Technology (Guangzhou), Research Assistant (2024 – 2025)

Research Interests

Financice Technology, Asset Pricing, NLP, ...

NOTES

数组 Array

链表 Linked List

哈希表 Hash Table

字符串 String

栈和队列 Stack & Queue

二叉树 Binary Tree

回溯算法 Back Tracking

贪心算法 Greedy Algorithm

动态规划 Dynamic Programming

单调栈 ​Monotonic Stack

图论 Graph Theory

Full-Stack Feed

Binary Brainery: Career Feed

Deconstruct algorithmic patterns into bite-sized, memorable chunks with our modular learning system. Each concept comes with video explainers or visual cheatsheets. Because understanding beats memorization – this is where coding literacy meets career acceleration.

  • Information Technology Basic Knowledge (High School's Teachers Level) notes videos

  • (随笔)贪心算法理论基础的通俗理解 ​Article Link​

  • Approaches to Interdisciplinary Research Paper Study(Shortlisted for THU/PKU Recommended Admission) Example 1 Example 2

  • Literature Review of LLM in Asset Pricing base on investors Research Analysis

CodeCanvas: LeetCode Contest

Sharpen your blade in the LeetCode Arena! We dissect weekly/biweekly contest problems through the lens of competitive programming – delivering multi solution, time complexity tradeoff analysis, and live leaderboard climbing tactics. Because every line of code here is a step toward your personal PB.

Prior to proceeding, it's advisable to revisit the foundational algorithms outlined below. Should time permit, consulting the notes section above may provide a more comprehensive reference.

基础算法 Basic Algorithms

Ctrl + Alt + Elite: Projects Hub

Reverse-engineer trending GitHub repos, then reimagine them with cutting-edge twists. This is a collaborative playground for architectural forensics – where we dissect tech stacks like digital surgeons. Coming soon: distributed system simulations, full-stack project DNA recombination, and open-source ethos survival kits.

  • Project 1 : The Impact of Company-Level Events on Stock Prices

    This is a financial research tool designed to analyze the impact of company-level events on stock prices. Its core functions include calculating abnormal returns (AR) and visualizing how specific events influence stock performance. The tool integrates statistical models to isolate event-driven price movements from market-wide trends, enabling researchers to quantify and graphically represent the causal relationship between corporate actions (e.g., earnings announcements, mergers) and equity valuation changes. (Sample Data & Code)

  • Project 2 : Attention Map

    The project aims to ​visualize attention mechanisms in Chinese texts, providing intuitive heatmaps to demonstrate the semantic relevance strength between different words when processed by Transformer models such as BERT.

 DEEPLEARNING

这部分的内容是基于《动手学深度学习v2》。在note部分,将教材中重要的概念和公式做了批注。在code部分,简化为仅用于实现章节概念的代码并修改调试跑通,添加了大量批注以配合初学者的理解;当然,您也可以在我的仓库中下载ipynb格式的代码文件。

线性神经网络 Linear Networks

  • Ch1 线性神经网络:线性回归 + 基础优化算法 note code

  • Ch1 线性神经网络:Softmax回归+损失函数+图像分类数据集 note code

多层感知机 Multilayer Perceptrons

  • Ch2 多层感知机:多层感知机 note code

  • Ch2 多层感知机:模型选择+欠拟合和过拟合 note code

  • Ch2 多层感知机:权重衰退 note code

  • Ch2 多层感知机:丢弃法 note code

  • Ch2 多层感知机:数值稳定性+模型初始化和激活函数 note code

卷积神经网络 Convolutional Neural Networks

  • Ch3 卷积神经网络:卷积 note code

  • Ch3 卷积神经网络:填充和步幅 note code

  • Ch3 卷积神经网络:多个输入输出通道 note code

  • Ch3 卷积神经网络:池化层 note code

  • Ch3 卷积神经网络:经典卷积神经网络LeNet note code

现代卷积神经网络 Modern Convolutional Neural Networks

  • Ch4 现代卷积神经网络:深度卷积神经网络AlexNet note code

  • Ch4 现代卷积神经网络:使用块的网络(VGG) note code

  • Ch4 现代卷积神经网络:网络中的网络(NiN) note code

  • Ch4 现代卷积神经网络:含并行连结的网络(GoogLeNet) note code

  • Ch4 现代卷积神经网络:批量规一化 note code

  • Ch4 现代卷积神经网络:残差网络(ResNet) note code

编译器和解释器 Computational Performance

  • Ch5 编译器和解释器:深度学习硬件:CPU、GPU、TPU、多卡并行 note code

  • Ch5 编译器和解释器:分布式计算 note

计算机视觉 Computer Vision

  • Ch6 计算机视觉:图像增广 note code

  • Ch6 计算机视觉:微调 note code

  • Ch6 计算机视觉:物体检测 note code

  • Ch6 计算机视觉:锚框 note code

  • Ch6 计算机视觉:物体检测算法——R-CNN+SSD+YOLO note code

  • Ch6 计算机视觉:语义分割 note code

  • Ch6 计算机视觉:转置卷积 note code

  • Ch6 计算机视觉:全连接卷积神经网络 note code

  • Ch6 计算机视觉:样式迁移 note code

循环神经网络 Recurrent Neural Networks

现代循环神经网络 Modern Recurrent Neural Networks

  • Ch8 现代循环神经网络:门控循环单元GRU note code

  • Ch8 现代循环神经网络:LSTM+深层+双向循环神经网络 note code

  • Ch8 现代循环神经网络:编码器-解码器 note code

  • Ch8 现代循环神经网络:序列到序列学习(seq2seq) note code

  • Ch8 现代循环神经网络:束搜索 note

注意力机制 Attention Mechanisms

  • Ch9 注意力机制:注意力机制+注意力分数+使用注意力机制的seq2seq note code

  • Ch9 注意力机制:自注意力和位置编码 note code

  • Ch9 注意力机制:Transformer note code

  • Ch9 注意力机制:BERT note code

  • Ch9 注意力机制:BERT微调 note code

  • Ch9 注意力机制:优化算法 note