Mining Incomplete Data—A Rough Set Approach - Semantic Scholar

Mining Incomplete Data—A Rough Set Approach Jerzy W. Grzymala-Busse [email protected]

Department of Electrical Engineering and Computer Science, University of Kansas, Lawrence, KS 66045, USA and Institute of Computer Science, Polish Academy of Sciences, 01-237 Warsaw, Poland

Mining Incomplete Data—A Rough Set Approach – p. 1/6

Sequential Methods, I Deleting cases with missing attribute values (listwise deletion, casewise deletion, complete case analysis) The most common value of an attribute The most common value of an attribute restricted to a concept Assigning all possible attribute values to a missing attribute value Assigning all possible attribute values restricted to a concept

Mining Incomplete Data—A Rough Set Approach – p. 2/6

Sequential Methods, II Replacing missing attribute values by the attribute mean Replacing missing attribute values by the attribute mean restricted to a concept Global closest fit Concept global fit Imputation ML method (maximum likelihood) EM method (expectation-maximization) Single random imputation Multiple random imputation

Mining Incomplete Data—A Rough Set Approach – p. 3/6

Parallel Methods C4.5 CART MLEM2 Characteristic Relations Singleton, Subset, and Concept Approximations Local Approximations Rule Induction

Mining Incomplete Data—A Rough Set Approach – p. 4/6

Incomplete Data Missing attribute values: Lost values are denoted by ? "do not care" conditions are denoted by * attribute-concept values are denoted by – All decision values are specified For each case at least one attribute value is specified

Mining Incomplete Data—A Rough Set Approach – p. 5/6

An Incomplete Decision Table Attributes

Decision

Case

Temperature

Headache

Nausea

Flu

1 2 3 4 5 6 7 8

high very_high ? high high normal normal –

– yes no yes ? yes no yes

no yes no yes yes no yes *

yes yes no yes no no no yes

Mining Incomplete Data—A Rough Set Approach – p. 6/6

Extendable Dialog Script Description Language for Natural Language User Interfaces 13/4/16

Yahoo Japan Research Kiyoshi Nitta [email protected] Copyright (C) 2013 Yahoo Japan Corporation. All Rights Reserved.

Table of Contents for INTELLI 2013 panel

P3

1. Artificial General Intelligence and Narrow AI 2. Rapidly Growing LOD Project Community 3. Breakthrough in AGI Side 4. Large-Scale Dialog Scripts: A Solution

Copyright (C) 2013 Yahoo Japan Corporation. All Rights Reserved.

Artificial General Intelligence and Narrow AI Artificial General Intelligence

Brain Emulation

P13

Narrow AI

Expert System Pragmatic Logic

Mind Emulation

5th Generation Computing (Japan) Artificial Life Social Intelligence

Human Made Knowledge Base (Cyc) Semantic Web

Natural Language Processing Speech Recognition Image Recognition Machine Learning Copyright (C) 2013 Yahoo Japan Corporation. All Rights Reserved.

Rapidly Growing LOD Project Community

P14

2009

(Bizer et al. 2009)

Linking Open Data cloud diagram, by Richard Cyganiak and Anja Jentzsch. http://lod-cloud.net/ Copyright (C) 2013 Yahoo Japan Corporation. All Rights Reserved.

Breakthrough in AGI Side

P15

Manually constructed knowledge base

Collaboratively accumulated knowledge base

Photo by inmymemory

Photo by inmymemory

Copyright (C) 2013 Yahoo Japan Corporation. All Rights Reserved.

Large-Scale Dialog Scripts: As An Application

Large-Scale Dialog Scripts

Large-Scale Knowledge Base

Information Retrieval

Watson

P16

Question Answering

WolframAlpha

Conversational Agents

Evi (TrueKnowledge)

Proactive Agents

Siri

Copyright (C) 2013 Yahoo Japan Corporation. All Rights Reserved.