Machine Learning Basics
Lecturer, School of Computer Science
Lecture Overview
Copyright By PowCoder代写 加微信 powcoder
§ What is machine learning?
§ Categories of machine learning
§ How does machine learning work? Supervised learning workflow
§ Machine learning algorithms
§ Model evaluation
What is machine learning?
What is machine learning?
What is machine learning?
What is machine learning?
”A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P, if its performance at tasks in T, as measured by P, improves with experience E.”
— , Professor at University
Examples of digits from theMNISTdatabase
§ TaskT:classifyinghandwritten digits from images
§ PerformancemeasureP: percentage of digits classified correctly
§ TrainingexperienceE:datasetof images of handwritten digits
Applications of machine learning
§ Email spam detection
§ Face detection and matching in smart phones
§ Stock predictions
§ Product recommendations (e.g. Netflix, amazon) § Sentiment analysis
§ Self-driving cars
§ Post office (e.g. sorting letter by post code)
§ Medical diagnoses
Categories of machine learning
§ Supervised learning Ø Labeled data
Ø Predict outcome/future
Classification predict categorical class labels
e.g. the handwritten digit (multi-class)
Regression
Prediction of continuous outcomes
e.g. students’ grade scores
Categories of machine learning
§ Unsupervised learning Ø No labels/targets
Ø Find hidden structure/insights in data
Dimensionality Reduction
– reduce data sparsity
– reduce computational cost
Clustering
Objectives within a cluster share a degree of similarity. e.g. product recommendation
Categories of machine learning
§ Reinforcement learning Ø Decision process
Ø Reward system
Ø Learn series of actions
Ø Applications: chess, video games, some robots, self-driving cars
Supervised learning workflow
hypothesis function h(x)
Source: Python machine learning, 2nd ed,
Supervised learning workflow
Source: Python machine learning, 2nd ed,
Some algorithms
Model evaluation – misclassification error
Model evaluation – other metrics
§ Accuracy (1-Error)
§ ROC, AUC
§ Precision, Recall
§ F-measure, G-mean
§ (Cross) Entropy
§ Likelihood
§ Squared Error/MSE
§ Major concepts of machine learning at a high level.
§ Different types of machine learning tasks.
§ The major steps of supervised learning: the workflow § Machine learning algorithms and evaluation
Learning Resources
§ Free machine learning eBooks https://github.com/rasbt/pattern_classification/blob/master/resources/machine_le arning_ebooks.md
§ 10-min Video by : https://www.youtube.com/watch?v=elojMnjn4kk&list=PL5- da3qGB5ICeMbQuqbbCOQWcS6OYBr5A&index=1
If you cannot access the video https://github.com/justmarkham/scikit-learn- videos/blob/master/01_machine_learning_intro.ipynb
§ Recommended reading:
Ø Raschka and Mirjalili: Python Machine Learning, 2nd ed., Ch 1
Ø , , and : The Elements of Statistical Learning, Ch 01 https://web.stanford.edu/~hastie/ElemStatLearn/printings/ESLII_print12.pdf
程序代写 CS代考 加微信: powcoder QQ: 1823890830 Email: powcoder@163.com