Presentation is loading. Please wait.

Presentation is loading. Please wait.

Situation decomposition method extracts partial data which contains some rules. Hiroshi Yamakawa (FUJITSU LABORATORIES LTD.)

Similar presentations


Presentation on theme: "Situation decomposition method extracts partial data which contains some rules. Hiroshi Yamakawa (FUJITSU LABORATORIES LTD.)"— Presentation transcript:

1 Situation decomposition method extracts partial data which contains some rules. Hiroshi Yamakawa (FUJITSU LABORATORIES LTD.)

2 Abstract In an infantile development process, the fundamental knowledge about the external world is acquired through learning without clear purposes. An adult is considered to use that fundamental knowledge for various works. The acquisition of the internal model in these early stages may exist as a background of the flexible high order function of human brain. However, research of such learning technology is not progressing to nowadays. The system can improves prediction ability and reusability in the lasting work by using the result of learning without clear purposes. Then, we proposed the situation decomposition technology which chooses the partial information which emphasizes the relation "another attribute value will also change if one attribute value changes." Situation decomposition technology is the technology of performing attribute selection and case selection simultaneously from the data structure from which each example constitutes an attribute vector. The newly introduced Matchability criteria are the amount of evaluations which becomes large, when the explanation range of the selected partial information becomes large and a strong relation exists in the inside. Processing of situation decomposition extracts plural partial situations (result of attribute selection and case selection) of corresponding to the local maximum points over this evaluation. Furthermore, extraction of partial problem space (based on the Markov decision process) is possible using the technology which extended situation decomposition in the direction of time. In action decision task, such as robot control, partial problem space can be assigned as each module of multi-module architecture. Then, it can be efficiently adapted to unknown problem space by combining extracted plural partial problem space.

3 My strategy for Brain-like processing Brain has very flexible learning ability. The intelligent processes which has more flexible learning abilities are more close to real brain processes. I want introduce learning ability to my system as possible.

4 Contents 1. Development and Autonomous Learning 2. SOIS (Self-organizing Information Selection) as Pre-task Learning 3. Delivering Matchable Principle 4. Situation Decomposition using Matchability Criterion 5. Application of Situation Decomposition 6. Conclusions & Future works

5 Autonomous Learning (Framework) Outline of this talk Pre-task learning Self-organizing Information Selection Situation decomposition Task learning Cognitive Development Situation Decomposition using Matchability Criterion Matchable Principle Matchability Criterion

6 Development and Autonomous Learning

7 Two aspects of Development “Acquired environmental knowledge without particular goals which helps for problem solving for particular goals”  → “Pre-task Learning” in Autonomous Learning “Calculation process which increases the predictable and/or operable object in the world”  → Enhancing prediction ability

8 Autonomous Learning: AL Two phases learning (Research in RWC) Task learningExisting Knowledge Acquiring environmental knowledge General fact For design Acquiring solution for goal goal No reaching over the wall Acquiring movable paths Generating path to the goal Environment is given Goal is given Pre-task learning Development Today’s Topic

9 Pre-task Learning helps Task Learning Autonomous Learning (AL)  Pre-task Learning Acquiring environmental knowledge without particular goal.  Task Learning Environmental knowledge speed up aacquiring solution for goal. In human:  Adult people can solve given task quickly using environmental knowledge acquired for other goal or without particular goal. Development ~ Pre-task Learning Development Today’s Topic

10 Research topics for AL Pre-task Learning (How to acquire environmental knowledge)  Situation Decomposition using Matchability criterion Situation Decomposition is kind of a Self-organizing Information Selection technology. Task learning (How to use environmental knowledge)  CITTA (Cognition based Intelligent Transaction Architecture) Multi-module architecture which can combining environmental knowledge acquired during Pre-task learning  Cognitive Distance Learning Goal driven problem solver for each environmental knowledge. Development Today’s Topic

11 Overview of Approaching for AL CITTA Combining environmental knowledge Situation Decomposition Acquiring environmental knowledge Cognitive Distance Learning Problem solver for each environmental knowledge Architecture Learning algorithm Pre-task Learning Task Learning

12 SOIS (Self-organizing Information Selection) as Pre-task Learning

13 SOIS: Self-organizing Information Selection Process: Selecting plural partial information from data.  → “Situation Decomposition” Criterion: Evaluation for each partial information.  → Matchability Criterion Knowledge = Set of structure. Partial Information = One kind of structure ※ SOIS could be a kind of knowledge acquiring process in development

14 Situation Decomposition is kind of SOIS For situation decomposition Partial Information = Situation Extracting plural situations which are combination of selected attributes and cases from spread sheet. MS 4 attributes Cases MS 1 MS 2 MS 3

15 Delivering Matchable Principle

16 Two aspects of Development “Acquired environmental knowledge without particular goals which helps solving problem for particular goals”  → “Pre-task Learning” in Autonomous Learning “Calculation process which increases the predictable and/or operable object in the world”  → Enhancing prediction ability

17 How to enhance prediction ability We needs Criterion for selecting situation.  We wants to extract local structures. Multiplex local structure is mixed in real world data MS 4 MS 1 MS 2 MS 3 Situation Decomposition

18 Deriving Matchable Principal What is Criterion for each selecting situation. Matchable principle  “S tructures where a matching opportunity is large are selected.” Extracting structure (knowledge) without particular goals. Prediction is based on matching a case with experiences.

19 Factors in Matchable Principle To increase matching opportunity Simplicity of Structure Ockham’s razor MDL 、 AIC Consistency for Data Coverage for Data Our proposed Matchability criterion Relation in Structure Accuracy Minimize error Case-increasing Attribute-increasing Association rule

20 SD (Situation Decomposition ) and Implementation

21 Situation Decomposition Extracting plural situations which are combination of selected attributes and cases from spread sheet. Matchability=This criteria evaluates matching opportunity Matchable Situation = Local maximums of Matchability MS 4 attributes Cases MS 1 MS 2 MS 3

22 Formalization: Whole situation and Partial situations Whole situation J=(D, N) : Contains N attributes and D cases. Attribute selection vector:  d = (d 1, d 2, …,d D ) Case selection vector :  n = (n 1, n 2, …,n N ) Vector element d i,n i are binary indicator of selection/unselection. Number of selected attributes: d Number of selected cases : n Situation decomposition extracts some matchable situations from whole situation J=(D, N) which potentially contains 2 D+N partial situation.

23 Case selection using Segment space Segment space is multiplication of separation of each selected attributes. (example: two dimension) n : Number of selected cases S d : Number of total segments r d : Number of selected segments ※ Cases inside the chosen segments are surely chosen. Sd =s1 s2Sd =s1 s2 attribute1 attribute2

24 [Number of selected cases] n →Make Larger [Number of total segments] Sd →Make Larger Matchability criterion from Matchable Principle nn SdSd rdrd rdrd N: Total number of cases, C 1, C 2, C 3 : Positive constant [Number of selected segments] rd →Make Smaller Simplicity of Structure Coverage for Data

25 Matchability Focuses in covariance Types of Relations  Coincidence The relation to which two events happen simultaneously  Covariance The relation that another attribute value will also change if one attribute value changes Matchability:  Estimates covariance in selected data for categorical attributes. ABC ⅰ8010 ⅱ 8010 ⅲ 80

26 How to find situations Algorithms searches local maximums of Matchability Criterion. Algorithm Overview  for each subset of d of D  Search Local maximums  Reject saddle point  end Time complexity ∝ 2 D

27 Simple example Input situation  Mixture of cases on two plains. Situation A: x + z = 1 Situation B: y +z = 1 Extracted situation  Input Situations MS 1= Input Situation A MS 2= Input Situation B  A New Situation MS 3 :  line x = y, x + z = 1

28 Generalization ability Multi-valued function φ:(x,y)→z Even if the input situation A (x+z=1) lacks half of its parts, such that no data exists in the range y>0.5, our method outputs φ MS1 (0,1)=1.0.

29 Applications of Situation Decomposition (SD)

30 Multi-module Prediction System InputOutput

31 ● Training cases 500 cases are sprayed on each plain in uniform distribution in the range x=[0.0, 1.0] and y=[0.0, 1.0]. ● Test cases 11×11 cases are arranged to notches at a regular interval of 0.1 on each plane Training cases and Test cases q: sampling rate

32 Prediction Result without Matchable Situation with Matchable Situation

33 Autonomous Learning: AL Two step learning (Research in RWC) Task learningExisting Knowledge Acquiring environmental knowledge General fact For design Acquiring solution for goal goal No reaching over the wall Acquiring movable paths Generating path to the goal Environment is given Goal is given Pre-task learning Development Today’s Topic

34 Demonstration of Autonomous Learning Door & Key task with CITTA Start Mobile Agent Door Telephone Key Goal Agent acquire knowledge as situation Door can open by the key.

35 Input/Output Each Situation is used as Module PositionActionObjectBelongings Matchable Situation i Matchable Situation 1 Go by wall Go straight Matchable Situation 2 Open door by telephone Open door by Key Extracting Matchable Situation Pre-task Learning Combining Matchable Situation Task Learning … Environment Mobile Agent

36 Situation Decomposition in AL SD in Pre-task learning:  Situation decomposition handles input /output vector of two time step for extracts Markov process. Advantages by SD in Task learning:  Adaptation by combining situations are possible.  Learning data can be reduced, because learning space for each module is reduced.

37 Conclusions and Future works

38 Autonomous Learning Conclusions Pre-task learning Self-organizing Information Selection Situation decomposition Task learning Cognitive Development Situation Decomposition using Matchability Criterion Matchable Principle Matchability Criterion

39 Conclusions & Future work Situation decomposition Matchability is new model selection criterion maximizing matching opportunity, which emphasize Coverage for data. In opposition ockham’s razor emphasize the Consistency for data. Decomposed situations by matchability criterion has powerful prediction ability. Situation decomposition method can be applied to pre-processing of data analysis, self-organization, pattern recognition and so on.

40 Future work Situation decomposition:  Needs theoretical research on Matchabilty criterion. This intuitively delivered criterion affected unbalanced data.  Needs speed up for large-scale problem. Exponential time complexity for number of attribute is awful.  Advanced Self-organized Information Selection Situation decomposition method only selects set of attributes and cases Autonomous Learning:  Relates with the knowledge of cognitive science.


Download ppt "Situation decomposition method extracts partial data which contains some rules. Hiroshi Yamakawa (FUJITSU LABORATORIES LTD.)"

Similar presentations


Ads by Google