- 博客(36)
- 收藏
- 关注
RSS订阅原创 Daily Tips
Tip : Download Source code from Github to local IntelliJ IDEA1. Copy the URL from Github, 2. Open IntelliJ, File -> New -> Project from Version Control3. Paste the URL in below field, then click Clone*** in the right botton corner, we can swithch to differ
2023-10-10 06:43:46
103
原创 随笔：Spring Test -- Mockito -- 浅谈个人理解@Mock, @MockBean, @InjectMocks
【代码】随笔：Spring Test -- Mockito -- 浅谈个人理解@Mock, @MockBean, @InjectMocks。
2023-04-10 18:04:37
568
1
原创 Computer Vision L9 -- Vision 3D
Why not using LiDAR3D sensor?It is very expensive It doesn't work in all conditions. In rainy days,laser will interact with a raindrop and not come back to the sensor, in that case we cannot build a complete 3D map.Machine Learning based approach...
2022-04-30 00:10:04
163
原创 Computer Vision L5.5 -- Interpretability of CNN for image detectio
Convolution networks right now seem to be black box to us. We want to know what are the networks learning specifically.When we construct the artificial neural net, we basically follow an analogy to the brain: neurons are kind of loosely based on the su
2022-04-28 08:45:59
124
原创 Computer Vision L8 -- Synthesis
We've been talking a lot about recognition and seeing, understanding the image. What we're talking about now is how to actually make a machine that creates an image.The basic idea behind this is that you actually have two different models that are
2022-04-21 14:49:12
92
原创 Computer Vision L7 -- Self-supervised Learning
How do we make the learning agent just consumes the data itself, when it doesn't get back to access to the labels.the main idea here is we have a dataset with no labels anymore. So what an encoder says is maybe I can just map it to itself.I can l...
2022-04-21 14:48:54
3095
原创 Computer Vision L6 -- Object Tracking
How do we track object over time?For example, the greener the car is saturated, it means the car is moving faster in the direction of down-left (reference is the color map on the right)However, our perception of motion is often an illusion, for examp
2022-04-19 16:08:52
2822
原创 Computer Vision L5 -- Video Recognition
Vedio would be a stack of images in time sequence, so we can still regard it as a function, but it would be in 3-dimensions.
2022-03-30 19:29:09
3034
原创 Computer Vision L4 -- Back Propagation
We can easily find result in forward pass
2022-03-28 20:23:27
64
原创 Computer Vision L5 -- Object Recognition
So the network firstly pick up random weights for each neuron, and output shows that Kitchen has the most probability. While the ground truth is that we, as humans,100% believe it is actually Bedroom. Now that we have preditions (blue chart) and ground t..
2022-03-08 19:59:58
93
原创 Computer Vision L3 -- Machine Learning Intro
What Machine Learning is aiming to do is figuring out what s should be such that function f would map the input x to y.The computer uses the loss function to find/optimizes. It tellshow far away our prediction is from what actually it should be. ...
2022-03-07 18:25:20
145
原创 Computer Vision L2 -- Fourier Transform
And we can do this Fourier Transform for a 2D image as well as we regard image as function.What is Fourier Transform going to do?It takes the original signal and tells us, for all the waves that could sum up to form this signal, what amplit...
2022-02-16 17:45:59
1524
原创 Computer Vision L1 -- Convolution
We know that images can be treated as 2d matrix. But we can also treat it as a function, where we give the location of a pixel (x,y) and get the intensity of that pixel.Now what we can do for denoising is Moving average. Imagine we have a imgae below...
2022-02-15 19:47:18
1589
原创 Multi-optimization (without using weights)
The big challenge in today's engineering field is that how we optimize something where the fitness criteria involves multiple factors and these factors are contradictory.(For example, when we build a bridge, we want it to be stable and light, these two..
2021-11-16 16:52:12
205
原创 Dimension reduction -- L6 for Data Science
Though some features have collinearity to each other, they still carry on some different informantion, so it would be wasteful to just drop one of these features. Can we do better?Yes, by dimension reduction, we can extract the independent information ..
2021-11-16 13:58:04
83
原创 Feature selection, regularization -- L5 for Data Science
*The observations are identically distributed means that we are just making random observations without any bias.* When we have different set of observations, we will get different value of point estimates of coefficients.For each case, the red st...
2021-11-10 01:54:18
85
原创 Review & Gradient Descent -- L4 for Data Science
Goal: We want to find a combination of betas such that it minimized the residual.The interpretatoin of the betas after we get them fitted:Quantify the relation between x_i and y. Like when we change a unit of x_1 in the set of x_i , what is the corr...
2021-11-09 18:07:15
83
原创 Evolutionary Algorithm--Chapter 3 Genetic Programming
Where are we now?When we say linear regression, we know that there is a linear mode. When saying non-linear regression, there is a non-linear model.A symbolic regression is a special case that we don't even know what the type offunction is at ...
2021-10-19 07:20:04
90
原创 Classification -- L3 for Data Science
Classification is a general extension of linear regression.0 means benign and 1 means malignant.In the graph, we can tell that if the average of radius is smaller than 10, it's very likely that the outcome will be benign. And if the radius...
2021-10-17 19:08:18
68
原创 Categorical Variables and One-Hot Encoding
Categorical VariablesCategorical variables are those which can take values from a finite set. For such variables, the values that they take up can have an intrinsic ordering (for e.g.,speed: {low, medium, high}). Such variables are called asOrdinal Cat..
2021-10-17 16:38:19
98
原创 Evolutionary Algorithm--Chapter 2 Genetic Algorithm
In beam search, when we rank the population, the rank for top 50% performance, thenthe bad solutions are abandoned, and good solutions could be duplicated and spawned off two searches from that point.And the difference between a beam search and a g...
2021-10-05 11:48:21
89
原创 Evolutionary Algorithm--Chapter 1 Parametric Optimization
Parametric optimization is the basis of evolution algorithm. The whole field of evolution algorithm could be seen as a special case ofoptimization, and parametric optimization is a particular situation that we have a function that has parameters. (Not a...
2021-10-05 05:06:32
95
原创 Linear Regression -- L2 for Data Science
Two questions here:1. why we use the four variables but not something else? 2. If we know these are four independent variables we are going to use, how do we determine the parameters in the equation ()?We want to make a prediction that i...
2021-10-04 08:13:19
106
原创 Statistics Primer-- L1 for Data Science
Basically Census is an effort to collect information about the population. But it is always impossible to get the information from every object in the population. So we need to draw some samples to get an inference. The sampling method then will be impor..
2021-10-02 14:02:59
88
原创 浅谈SVD
SVD, Single Value Decomposition , is essentially a waythat one can factorize a matrix.When we talk about this methodin aspect of scalar like 10, we canfactorize it by 2*5, or 30 by 2*3*5.And when we factorize a matrix, we can write it in a spec...
2021-10-02 11:41:54
84
原创 Artificial Intelligence -- Chapter 12 Intro to Machine Learning
One thing to point out: the same feature can appear multiple times in a decision tree as long as the're on separate paths. (Look at the feature "Alternate" in this tree.)Purer means after the attributefiltering, we should be more clear ...
2021-06-13 01:53:57
1163
1
原创 Artificial Intelligence -- chapter 11 Inference and Sampling
The task is to tranfer from the CPT we have to the CPT we want -- P( X | e )There are many factors that have influence on the Query variables (the information we want), but we only have access to some of those factors (blue evidence), other factors tha..
2021-06-12 05:00:07
79
原创 Artificial Intelligence -- chapter 10 Bayesian Networks
Bayesian networks is sort of a generalization of the Hidden Markov model. Instead of having a really rigid structure of state evidence. Here in Bayesian networkswe will table about arbitrary relationships variables, independences among them and what we ...
2021-06-06 07:08:35
87
原创 Artificial Intelligence -- Chapter 9 Hidden Markov Models
We can't observe states, but we can observe some indirect evidence about the states, and we will use these evidence to infer what the state is like. (We see E, and we infer X by E)Now we have an observation model, this is the information about what we ..
2021-06-05 22:24:04
142
2
原创 Artificial Intelligence -- Chapter 8 Probability and Markov Chains
The big idea is that we will still be keeping track of the information like states and such, but in addition to that, we will also keep trackof our certainty, numerically speaking that means probabilities.The capital letter represents for Randon Va..
2021-06-04 05:46:49
65
原创 Artificial Intelligence - Chapter 7 Reinforcement Learning
Why reinforcement learning?When we do dynamic programming, we assume that we know the transition function (the model )andreward function. But in the real world, we often don't have access to the true models and information for reward function. Even if...
2021-05-31 04:35:02
395
原创 Artificial Intelligence - Chapter 5&6 Sequential Decision Problems & Dynamic Programming
In this chapter, we'll get back to single-agent problems and talk about itsdecision-making problems more generally in the aspect of stochastics.Welcome to the realm of uncertainty!Multi-armed bandit problems: We have many slot machines, each of the..
2021-05-27 03:05:09
92
原创 Artificial Intelligence - Chapter 3 # Local Search
In this chaprer, we're going to talking at a high level about a few different algorithms that are commonly used for local search.We talked about a sort of pathfinding problems. But sometimes we don't need the full path to solve the problem, we don't ne
2021-05-25 20:22:10
59
原创 Artificial Intelligence - Chapter 2 Search Problems
Seaching problems could be implemented by problem-solving agent, it is a subset of goal-based agents. The goal here is a certain state. So theses agents will want to get to a goal state through a combination of actions and to do that they're going to sea..
2021-05-16 23:35:18
427
原创 Artificial Intelligence - Chapter 1 Introduction
人工智能 - 序关于人工智能，主要是两个维度的考虑：thinking & acting, 对于每个维度， 我们又可以考虑他们以何种方式执行，即humanly or rantionaly。 这样就产生了四种组合：1. Acting humanlyWhat is Turing test: a computer or a system is put behind the screnn and a human being is put behind another screen. You .
2021-05-15 00:16:57
130
空空如也
空空如也
TA创建的收藏夹 TA关注的收藏夹
TA关注的人