1 / 55

Machine Learning and Bioinformatics 機器學習與生物資訊學

Machine Learning and Bioinformatics 機器學習與生物資訊學. Enhancing instance-based classification using kernel density estimation. Outline. Kernel approximation Kernel a pproximation in O ( n ) Multi-dimensional kernel approximation Kernel d ensity e stimation

matia
Download Presentation

Machine Learning and Bioinformatics 機器學習與生物資訊學

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Machine Learning and Bioinformatics機器學習與生物資訊學 Machine Learning & Bioinformatics

  2. Enhancing instance-based classification using kernel density estimation Machine Learning and Bioinformatics

  3. Outline • Kernel approximation • Kernel approximation in O(n) • Multi-dimensional kernel approximation • Kernel density estimation • Application in data classification Machine Learning and Bioinformatics

  4. Identifying class boundary Machine Learning and Bioinformatics

  5. Boundary identified Machine Learning and Bioinformatics

  6. Kernel approximation Machine Learning and Bioinformatics

  7. Kernel approximationProblem definition • Given the values of function at a set of samples • We want to find a set of symmetric kernel functions and the corresponding weights such that Machine Learning and Bioinformatics

  8. Equations • (1) Machine Learning and Bioinformatics

  9. Kernel approximationAn 1-D example Machine Learning and Bioinformatics

  10. Kernel approximationSpherical Gaussian functions • Hartman et al. showed that a linear combination of spherical Gaussian functions can approximate any function with arbitrarily small error • “Layered neural networks with Gaussian hidden units as universal approximations”, Neural Computation, Vol. 2, No. 2, 1990 • With the Gaussian kernel functions, we want to find such that Machine Learning and Bioinformatics

  11. Equations • (1) • (2) Machine Learning and Bioinformatics

  12. Kernel approximation in O(n) Machine Learning and Bioinformatics

  13. Kernel approximation in O(n) • In the learning algorithm, we assume uniform sampling • namely, samples are located at the crosses of an evenly-spaced grid in the d-dimensional vector space • Let denote the distance between two adjacent samples • If the assumption of uniform sampling does not hold, then some sort of interpolation can be conducted to obtain the approximate function values at the crosses of the grid Machine Learning and Bioinformatics

  14. A 2-D example of uniform sampling Machine Learning and Bioinformatics

  15. The O(n)kernel approximation • Under the assumption that the sampling density is sufficiently high, i.e., we have the function values at a sample and its k nearest samples, , are virtually equal: • In other words, is virtually a constant function equal to in the proximity of • Accordingly, we can expect that ; Machine Learning and Bioinformatics

  16. The O(n)kernel approximationAn 1-D example Machine Learning and Bioinformatics

  17. Under the assumption that , we have • The issue now is to find appropriate and such thatin the proximity of Machine Learning and Bioinformatics

  18. Equations • (1) • (2) • (3) Machine Learning and Bioinformatics

  19. What’s the difference between equations (2) and (3)? Machine Learning and Bioinformatics

  20. Machine Learning and Bioinformatics

  21. Therefore, with , we can set and obtain • In fact, it can be shown that is bounded by • Therefore, we have the following function approximate: Machine Learning and Bioinformatics

  22. Equations • (1) • (2) • (3) • (4) • (5) Machine Learning and Bioinformatics

  23. The O(n)kernel approximationGeneralization • We can generalize the result by setting , where is a real number • The table shows the bounds of Machine Learning and Bioinformatics

  24. Equations • (1) • (2) • (3) • (4) • (5) • (6) Machine Learning and Bioinformatics

  25. The effect of Machine Learning and Bioinformatics

  26. The smoothing effect • The kernel approximation function is actually a weighted average of the sampled function values • Selecting a larger value implies that the smoothing effect will be more significant • Our suggestion is set Machine Learning and Bioinformatics

  27. An example of the smoothing effect Machine Learning and Bioinformatics

  28. Multi-dimensional kernel approximation Machine Learning and Bioinformatics

  29. Multi-dimensional kernel approximationThe basic concepts • Under the assumption that the sampling density is sufficiently high, i.e., we have the function values at a sample and its k nearest samples, , are virtually equal: • As a result, we can expect that and , where and are the weights and bandwidths of the Gaussian functions located at , respectively Machine Learning and Bioinformatics

  30. Let , the scalar form of is • If we set to , then is bounded by Machine Learning and Bioinformatics

  31. Accordingly, we have is bounded by and, with , • In RVKDE, d is relaxed to a parameter, • Finally, by setting uniformly to , we obtain the following function that approximates :, where are the k nearest samples of the object located at v Machine Learning and Bioinformatics

  32. Equations • (1) • (2) • (3) • (4) • (5) • (6) • (7) Machine Learning and Bioinformatics

  33. Multi-dimensional kernel approximationGeneralization • If we set uniformly to, then • As a result, we want to set , where and obtain the following function approximate:,where are the k nearest samples of the object located at v • In RVKDE, k is a parameter kt Machine Learning and Bioinformatics

  34. Equations • (1) • (2) • (3) • (4) • (5) • (6) • (7) • (8) Machine Learning and Bioinformatics

  35. Multi-dimensional kernel approximationSimplification • An interesting observation is that, as long as , then regardless of the value of , we have • With this observation, we have Machine Learning and Bioinformatics

  36. Equations • (1) • (2) • (5) • (6) • (7) • (8) • (9) Machine Learning and Bioinformatics

  37. Machine Learning and Bioinformatics

  38. Lack something? ? Machine Learning and Bioinformatics

  39. Kernel density estimation Machine Learning and Bioinformatics

  40. Kernel density estimationProblem definition • Assume that we are given a set of samples taken from a probability distribution in a d-dimensional vector space • The problem definition of kernel density estimation is to find a linear combination of kernel functions that provides a good approximation of the probability density function Machine Learning and Bioinformatics

  41. Probability density estimation • The value of the probability density function at a vector v can be estimated as follows: • n is the total number of samples • R(v) is the distance between vector v and its ks-thnearest samples • is the volume of a sphere with radius = R(v) in a d-dimensional vector space Machine Learning and Bioinformatics

  42. The Gamma function • , since • , since Machine Learning and Bioinformatics

  43. Application in data classification Machine Learning and Bioinformatics

  44. Application in data classification • Recent development in data classification focuses on the support vector machines (SVM), due to accuracy concern • RVKDE • delivers the same level of accuracy as the SVM and enjoys some advantages • constructs one approximate probability density function for one class of objects Machine Learning and Bioinformatics

  45. Time complexity • The average time complexity to construct a classifier is , if the k-d tree structure is employed, where n is the number of training samples • The time complexity to classify m new objects with unknown class is Machine Learning and Bioinformatics

  46. Comparison of classification accuracy Machine Learning and Bioinformatics

  47. Comparison of classification accuracy3 larger data sets Machine Learning and Bioinformatics

  48. Data reduction • As the learning algorithm is instance-based, removal of redundant training samples will lower the complexity of the prediction phase • The effect of a naïve data reduction mechanism was studied • The naïve mechanism removes a training sample, if all of its 10 nearest samples belong to the same class as this particular sample Machine Learning and Bioinformatics

  49. With the KDE based statistical learning algorithm, each training sample is associated with a kernel function with typically a varying width Machine Learning and Bioinformatics

  50. In comparison with the decision function of SVM in which only support vectors are associated with a kernel function with a fixed width Machine Learning and Bioinformatics

More Related