CSTalks - On machine learning - 2 Mar
Download as PPT, PDF0 likes420 views
This document summarizes machine learning concepts including supervised and unsupervised learning techniques. It discusses fundamental questions in machine learning like how to build systems that improve with experience. Key problems covered include classification, regression, and clustering. Challenges like overfitting, model complexity, and optimization techniques like gradient descent are also summarized. Open problems in machine learning like transferring learned knowledge between tasks and preserving privacy in data mining are mentioned.
CSTalks - On machine learning - 2 Mar
- 1. On Machine Learning at CSTalks by Vlad Hosu
- 2. Introduction
- 3. Fundamental Questions What are the fundamental laws that govern all learning processes? How can we build computer systems that automatically improve with experience?
- 4. Learning: Method a process of adaption by which a parametric model is automatically adjusted so that some fitness criteria is more readily met
- 5. Before Learning I’m learning, hence I need adapt!
- 6. After Result: Liony adjusts his diet.
- 7. Biological Learning Model: nervous system neuron connectivity, chemical changes etc Fitness: improved behavior skills, memory, knowledge
- 8. Machine Learning a mathematical model with adjustable parameters optimizing some fitness function
- 9. Motivation
- 10. Why? some things are hard to code too much data automatic learning works better is easier to customize/personalize
- 11. Learning: Purpose estimation function - stock market class - recognition structure - grouping
- 12. Requirements good learning ability scalability to large problems simple and easy algorithm implementation
- 13. Things Ahead Problems Clustering Classification Regression Learning issues importance of domain knowledge learning/generalization ability model complexity issues Optimization
- 15. Clustering
- 17. Classification Types discriminative generative x1 x2
- 18. Classification Types discriminative generative x1 x2 1 0
- 19. Classification
- 20. Regression
- 21. Making Connections discrete value regression => generative classification regression on boundary space => discriminative classification clustering + labels => classification
- 22. Learning Issues
- 23. Domain Knowledge exploitation of problem structure human abstractions are better important for picking the right model
- 24. Grouping in Images groups together similar parts of an image select objects find patterns features = pixel values (function of)
- 25. Segmentation
- 26. Color Space RGB space RGB space
- 29. Generalization Ability training data generalizes to new data important for classification accuracy
- 30. Support Vector Machines (SVM) linear classifier on distorted space
- 31. Learning Ability over fitting
- 32. Problems with Over-fitting
- 33. SVM vs Decision Trees
- 34. Complexity Issues models should be as simple as possible but representative of the training data
- 35. Neural Networks model: weights fitness: output error general function ∑
- 38. Optimization
- 39. Optimizing Fitness find extrema strategies gradient descent convex optimization
- 40. Optimization finding extrema local/global
- 41. Gradient Descent
- 43. Problem: Speed
- 44. Linear Programming x1 x2 lines define a convex function planes in 3D etc
- 45. Considerations scaling to large features spaces feature selection dimensionality reduction
- 46. Open Problems
- 47. Open Problems unlabeled data for regression exploiting sparsity in high dimensional spaces for non-parametric learning transferring learnt information from one task to simplify learning another
- 48. Open Problems (cont) algorithms for learning control strategies from delayed rewards and other inputs best “active learning” strategies for different learning problems degree one can preserve data privacy while obtaining the benefits of data mining
- 50. Types of Regression parametric non-parametric
- 51. Linear vs Non-linear linear smooth under-fitting good enough for some processes (biz) non-linear complex over-fitting works on most data-sets
- 52. Naive Bayes good spam write people free π π No. Good No. Spam * *
- 53. Graph Clustering
- 54. Mean Shift
- 55. Problems in CV What are the physical and geometric processes that govern (digital) imaging? What are the “informative” areas of an image and how do we detect them? What portions of an image pertain to one another and to relevant physical phenomena? From one (or more) images, how can we determine the geometry of the scene?
- 56. Linear Regression model: straight line 2 adjustable parameters fitness function: root mean squared error
- 57. Solution Stability y-shift slope
- 58. Some Issues with Model Selection normal outliers wrong model
- 59. Real Photo in Color Space EM KMeans
- 61. Newton’s Method