lzj509649444/machine-learning

Study Machine Learning in the Coursera

machine-learning-coursera

Coursera machine learning course resources.

Text book:

Bayesian Reasoning and Machine Learning http://web4.cs.ucl.ac.uk/staff/D.Barber/textbook/090310.pdf

Video lectures:

https://class.coursera.org/ml/lecture/preview

Schedule:

Week 1 - Due 07/04: DONE

• Introduction
• Linear regression with one variable
• Linear Algebra review (Optional)

Week 2 - Due 07/11: DONE

• Linear regression with multiple variables

• Octave tutorial

• Programming Exercise 1: Linear Regression

    Best and Most Recent Submission
    Score
    100 / 100 points earned PASSED
    Submitted on 6 七月 2015 在 7:35 晚上
    Part	Name	Score
    1	Warm up exercise	10 / 10
    2	Compute cost for one variable	40 / 40
    3	Gradient descent for one variable	50 / 50
    4	Feature normalization	0 / 0
    5	Compute cost for multiple variables	0 / 0
    6	Gradient descent for multiple variables	0 / 0
    7	Normal equations	0 / 0
  

Week 3 - Due 07/18: DONE

• Logistic regression

• Regularization

• Programming Exercise 2: Logistic Regression

    Best and Most Recent Submission
    Score
    100 / 100 points earned PASSED
    Submitted on 8 七月 2015 在 1:00 凌晨
    Part	Name	Score
    1	Sigmoid function	5 / 5
    2	Compute cost for logistic regression	30 / 30
    3	Gradient for logistic regression	30 / 30
    4	Predict function	5 / 5
    5	Compute cost for regularized LR	15 / 15
    6	Gradient for regularized LR	15 / 15
  

Week 4 - Due 07/25: DONE

• Neural Networks: Representation

• Programming Exercise 3: Multi-class Classification and Neural Networks

    Best and Most Recent Submission
    Score
    100 / 100 points earned PASSED
    Submitted on 9 七月 2015 在 1:16 凌晨
    Part	Name	Score
    1	Regularized logistic regression	30 / 30
    2	One-vs-all classifier training	20 / 20
    3	One-vs-all classifier prediction	20 / 20
    4	Neural network prediction function	30 / 30
  

Week 5 - Due 08/01: DONE

• Neural Networks: Learning

• Programming Exercise 4: Neural Networks Learning

    Best and Most Recent Submission
    Score
    100 / 100 points earned PASSED
    Submitted on 9 七月 2015 在 7:25 晚上
    Part	Name	Score
    1	Feedforward and cost function	30 / 30
    2	Regularized cost function	15 / 15
    3	Sigmoid gradient	5 / 5
    4	Neural net gradient function (backpropagation)	40 / 40
    5	Regularized gradient	10 / 10
  

Week 6 - Due 08/08: DONE

• Advice for applying machine learning

• Machine learning system design

• Programming Exercise 5: Regularized Linear Regression and Bias v.s. Variance

    Best and Most Recent Submission
    Score
    100 / 100 points earned PASSED
    Submitted on 11 七月 2015 在 3:28 凌晨
    Part	Name	Score
    1	Regularized linear regression cost function	25 / 25
    2	Regularized linear regression gradient	25 / 25
    3	Learning curve	20 / 20
    4	Polynomial feature mapping	10 / 10
    5	Cross validation curve	20 / 20
  

Week 7 - Due 08/15: DONE

• Support vector machines

• Programming Exercise 6: Support Vector Machines

    Best and Most Recent Submission
    Score
    100 / 100 points earned PASSED
    Submitted on 12 七月 2015 在 2:48 凌晨
    Part	Name	Score
    1	Gaussian kernel	25 / 25
    2	Parameters (C, sigma) for dataset 3	25 / 25
    3	Email preprocessing	25 / 25
    4	Email feature extraction	25 / 25
  

Week 8 - Due 08/22: DONE

• Clustering

• Dimensionality reduction

• Programming Exercise 7: K-means Clustering and Principal Component Analysis

    Best and Most Recent Submission
    Score
    100 / 100 points earned PASSED
    Submitted on 13 七月 2015 在 2:45 凌晨
    Part	Name	Score
    1	Find closest centroids	30 / 30
    2	Compute centroid means	30 / 30
    3	PCA	20 / 20
    4	Project data	10 / 10
    5	Recover data	10 / 10
  

Week 9 - Due 08/29: DONE

• Anomaly Detection

• Recommender Systems

• Programming Exercise 8: Anomaly Detection and Recommender Systems

    Best and Most Recent Submission
    Score
    100 / 100 points earned PASSED
    Submitted on 14 七月 2015 在 8:12 晚上
    Part	Name	Score
    1	Estimate gaussian parameters	15 / 15
    2	Select threshold	15 / 15
    3	Collaborative filtering cost	20 / 20
    4	Collaborative filtering gradient	30 / 30
    5	Regularized cost	10 / 10
    6	Gradient with regularization	10 / 10
  

Week 10/11 - Due 09/05: DONE

• Large scale machine learning
• Application example: Photo OCR

###Final Grade: 100%

Summary

-Supervised Learning

	Linear regression, logistic regression, neural networks, SVMs

-Unsupervised Learning

	K-means, PCA, Anomaly detection

-Special applications/special topics

	Recommender systems, large scale machine learning

-Advice on building a machine learning system

	Bias/variance, regularization; deciding what to work on next: evalution of learning algorithms, learning curves, error analysis, ceiling analysis.