Data Mining: Practical Machine Learning Techniques

Transcription

1 Artificial Intelligence Data Mining: Practical Machine Learning Techniques Dae-Won Kim School of Computer Science & Engineering Chung-Ang University

2 AI Scope 1. Search-based optimization techniques for real-life problems Hill climbing, Branch and bound, A*, Greedy algorithm Simulated annealing, Tabu search, Genetic algorithm 2. Machine Learning/ Pattern Recognition/ Data Mining Classification: Bayesian algorithm, Nearest-neighbor algorithm, Neural network Clustering: Hierarchical algorithm, K-Means algorithm 3. Reasoning: Logic, Inference, and knowledge representation Logical language: Syntax and Semantics Inference algorithm: Forward/Backward chaining, Resolution, and Expert System 4. Uncertainty based on Probability theory 5. Planning, Scheduling, Robotics, and Industry Automation

3 Did you ever hear about Big Data?

4 Progress in digital data acquisition and storage technology has resulted in the growth of huge databases.

5 Data mining is the extraction of implicit, previously unknown, and potentially useful information from data.

6 We build algorithms that sift through databases automatically, seeking patterns.

7 Strong patterns, if found, will likely generalize to make accurate predictions on future data.

8 Algorithms need to be robust enough to cope with imperfect data and to extract patterns that are inexact useful.

9 Machine learning provides the technical basis of data mining.

10 We will study simple machine learning methods, looking for patterns in data.

11 People has been seeking patterns in data since human life began.

12 In data mining, computer algorithm is solving problems by analyzing data in databases.

13 Data mining is defined as the process of (knowledge) discovering patterns in data.

14 Data mining is defined as the process of (knowledge) discovering patterns in data.

15 We start with a simple example.

16 Q: Tell me the name of this fish.

17 Algorithm??

18 We have 100 fishes, and measured their lengths. (e.g., fish: x=[length] t )

19

20 Our algorithm can measure the length of a new fish, and estimate its label.

21 Yes, it is a typical prediction task through classification technique. But, it is often inexact and unsatisfactory.

22 Next, we measured their lightness. (e.g., fish: x=[lightness])

23

24 Lightness is better than length.

25 Let us use both lightness and width. (e.g., fish: x=[lightness, width])

26

27 Each fish is represented a point (vector) in 2D x-y coordinate space.

28 Everything is represented as N- dimensional vector in coordinate space.

29 The world is represented as matrix

30 We assume that you have learned the basic concepts of linear algebra.

31 The objective is to find a line that effectively separates two groups.

32 How to find the line using a simple Math from high school?

33 We can build a complex nonlinear line to provide exact separation.

34

35 The formal procedure is given as:

36

37 This shows a predictive task of data mining, often called as pattern classification/ recognition/ prediction.

38 The act of taking in raw data and making an action based on the category of the pattern.

39 We build a machine that can recognize or predict patterns.

40

41 Another famous task of data mining is a descriptive task. Cluster analysis is the well-known group discovery algorithm.

42

43 We will experience the basic issues in the prediction task (pattern classification) in forthcoming weeks.

44 Some terms should be defined.

45 Given training data set : n x d pattern/data matrix: Fish Lightness Length Weight Width Class Label Fish Salmon Fish Salmon Fish Sea bass Fish Sea bass

46 Given training data set : n x d pattern/data matrix: d features (attributes, variables, dimensions, fields) Fish Lightness Length Weight Width Class Label Fish Salmon Fish Salmon Fish Sea bass Fish Sea bass n patterns (objects, observations, vectors, records)

47 Each pattern is represented as a feature vector.

48 The training pattern matrix is stored in a file or database.

49 Given labeled training patterns, the class groups are known a priori.

50 We constructs algorithms to classify new data into the known groups.

51 Training data vs. Test data

52 Training data are used as answers. We are learning algorithms using training data.

53 Test data are a set of new unseen data. We predict class labels using the learned algorithm.

54 Training data # of data # of features data index feature-1 feature-2 feature-n class label data index feature-1 feature-2 feature-n class label data index Feature-1 feature-2 feature-n class label Test data # of data # of features data index feature-1 feature-2 feature-n data index feature-1 feature-2 feature-n data index Feature-1 feature-2 feature-n

55 For example, we try to classify the tumor type of breast cancer patients

56 Breast-cancer-training.txt Patient cancer Patient normal Patient cancer Breast-cancer-test.txt Patient Patient Patient

57 To evaluate the performance of prediction algorithms, we need a performance measure (Accuracy).

58 Gold Standard (Truth) Prediction Result Positive Negative Positive True Positive False Positive Negative False Negative True Negative Suspicious Patients with Breast Cancer Prediction Result Positive (Cancer) Negative (Normal) Positive (Cancer) True Positive False Positive Negative (Normal) False Negative True Negative Accuracy = (True Positive + True Negative) / (True Positive + False Positive + False Negative + True Negative)

59 Gold Standard (Truth) Prediction Result Positive Negative Positive True Positive False Positive Negative False Negative True Negative Suspicious Patients with Breast Cancer Prediction Result Positive (Cancer) Negative (Normal) Positive (Cancer) 30 5 Negative (Normal) Accuracy = ( ) / ( ) = 0.85 (85%)