Olives_Blog

Quite Start

---

Project Target

In this project, we focus on the human face identified. And in this part, we use three different Nets to achieve our goal.

This project based on MTCNN to make our job works.

Quite Start

Background

The problem of the quickest descending line or shortcut is the first variational problem to appear in history and is a sign of the development of variational methods. The problem was proposed by John Bernoulli in 1696 in an open letter to his brother Jacob Bernoulli. The formulation of the problem is: let A and B are not in the same lead plane on the two points, in all the plane curve connecting A and B, to find a curve, so that only by gravity and the initial speed of the zero point from point A to point B along the curve of the shortest time required to move!

---

Demand

Using DP ways to solve this problem. Discrete this as grids, from the end to the start, calculate the faastest ways needs shortest times.

---

Mathematical Model

Let the curve be y=y(x) with coordinates as shown in Fig. 1, the mass starts moving from point O, its velocity v is related to its longitudinal coordinate:

$v=\sqrt{2gy}$
At the point (x, y) on the curve, the velocity of the mass point is

$v=\frac{ds}{dt}=\sqrt{1+y^{\mathrm{\prime }2}}\frac{dx}{dt}$
where s denotes the arc length of the curve and t denotes the time.

Background

Introduction

本部分用于记录深度强化学习中基于策略的学习使用的梯度，用于Critic、Actor 构成的状态评估函数的梯度求解。

---

Quite Start

---

Main part

十分简单的实现原理，参考背后数学推导即可完成最核心的loss部分。
另：CNN使用一维卷积，出了点bug，还没调完，不想做了，跳过task3

项目背景：完成实验课逻辑回归与支持向量机的机器学习实现
Based on: Python 3.9, pytorch in conda

---

项目原理

基于融合学习，进行游戏公司玩家流失量预测

基本信息

书名：《人人都是产品经理》
阅读日期：2024/03/02-2024/03/12
阅读次数： 1

---

第一次阅读笔记

---

DeepReinforceLearning

主要介绍深度强化学习中的DQn、梯度策略与AlphaGo及AlphaGo Zero

项目背景：出于需求微博部分大V提问信息、问题价值、问题回复以及对应回答人员的信息的爬取用于科研，不涉及反反爬内容，仅用最基本功能等。
Based on: Python 3.8

---

Welcome to Hexo! This is your very first post. Check documentation for more info. If you get any problems when using Hexo, you can find the answer in troubleshooting or you can ask me on GitHub.