[K Engine] Intorduction To Machine Planning, A Case Based Approach

Post on 24-May-2015

206 views 3 download

Tags:

description

My presentation to KEngine company about machine planning

transcript

Introduction to Machine Planning

Case Based Approach

Agenda

Case-Based Reasoning (CBR) Case-Based Planning (CBP)

What is Case-Based Reasoning?

CBR is a problem solving paradigm that instead of relying on general knowledge about the problem domain, CBR is able to utilize cases.

What is Case-Based Reasoning?

CBR is an approach to incremental learning, since new experience is retained each time a problem has been solved, making it available for future problems.

What is a Case?

Case is a specific knowledge of

previously experienced concrete problem situation which has been captured and learned in a way that it can be reused in solving future problems.

What is a Case?

New problem to be solved is called new case, or unsolved case.

Solved CaseUnsolved

CaseSolv

e

Previous experience is known as retained case, solved case, past case.

Case-based Problem Solving

Reasoning by re-using past cases is powerful and frequently applied way to solve problems for humans.

Case-based Problem Solving

It has been shown that human uses past-cases as models when learning to solve problems in early learning.

Learning in CBR

CBR denotes a machine learning paradigm that enables

sustained and incremental learning by updating the case-base after a problem has been solved.

Learning in CBR

Effective learning in CBR requires a well worked methods in order to:

1. Extract relevant knowledge from the experience.

2. Integrate a case into an existing knowledge structure.

3. Index the case for later matching with a similar case.

Fundamentals of case-based reasoning methods

Central tasks to all CBR methods:

1. Identify the current problem situation.

2. Find a past case similar to the new one.

3. Suggest a solution to the current problem.

4. Evaluate the proposed solution.

5. Update the system by learning from this experience.

CBR Cycle - Retrieve

1.

Retrieve the most similar case or cases.

CBR Cycle - Reuse

2. Reuse the information and knowledge in that case to solve the problem.

CBR Cycle - Revise

3. Revise the proposed solution.

CBR Cycle - Retain

4. Retain the parts of this experience likely to be useful for future problem solving.

CBR Cycle- Retrieve under the hood

1. Retrieve Identify Feature Search Initially Match Select

CBR Cycle – Reuse under the hood

2. Reuse Copy or Adapt Transformational VS Derivational

Reuse.

CBR Cycle – Revise under the hood

3. Revise Evaluate Solution. Repair Fault.

CBR Cycle – Retain under the hood

4. Retain Extract Index Integrate

CBR Tasks Decomposition

Problems in Case Representation

1. Deciding what to store in a case?

2. Finding an appropriate structure for describing the case content.

3. Deciding how the case memory (case-base) should be organized and indexed for fast retrieval and reuse?

4. How to integrate the case memory into a model of general domain knowledge.

What are CBR Challenges?

Core problems address by CBR research can be grouped into 5 areas, a solution to them constitute a CBR method:1. Knowledge/Case representation.2. Retrieval methods.3. Reuse methods.4. Revise methods.5. Retain methods.

What are CBR Challenges?

The BIG challenge is to come up with methods that are suited for problem solving and learning in a subject domain (e.g Medical), and for a particular application environments (e.g Heart failure diagnosis).

When an architect starts a new design.

When a surgeon approaches an operation.

When you get to your car to go home.

We use CBP everyday.

What is Case Based Planning?

CBP is the idea of planning as

remembering.Remembering success so they can be

reused.Remembering failure so they can be

avoided.Remembering repairs so they can be

applied.

CBP view to planning

The theory of case-based planning sees planning as an activity that tests a planner`s understanding of the world.

Planning errors lead the planner to learn more about its causes, which gives it a better understanding of how to avoid them.

Planning theories basic features

Case-based planner searches its memory for plans that satisfy many of its goals at once.

Greedy

Planning theories basic features

Case-based planner treats errors as opportunities to learn more about its domain, and the problems that arise in it.

Positive thinker

Planning theories basic features

Case-based planner saves previous built plan in memory for later use in similar circumstances.

Skimpy

Building Initial Plan

Planner stores and organizes the past planning experience under two sort of indices:

1. Goals to be satisfied

2. Failures to be avoided.

Why ?

Building Initial Plan

The case-based approach to finding an initial plan is to anticipate problems so the planner can find plans that avoid them

Debugging Failed Plan (F5)

Expectation failure VS Planning failure.

How the planner responds to each of them? Planning failure -> Replanning. Expectation failure -> Repair faulty

knowledge.

Learning from planning

A case-based planner learns 31. Plan learning: learn new plans that

avoid problems.

2. Expectation learning: learn features that predict the problems.

3. Critic learning: learn the repairs that have to be made if those problems arise again in different circumstances

Why case-based?

We want a planner that can learn

and recall complex plans rather than having to repeat work it has already done.

Rule based VS Case based needs.

Retriever

To retrieve plans from memory, the plan Retriever needs

1. A memory of plans indexed by the goals they satisfy.

2. A goal similarity metric.

3. A value hierarchy of goals.

Modifier

To alter old plans to meet new goals, Modifier needs

1. A set of modification rules.

2. Critics.

3. Knowledge of goal specific requirements and general plan specification.

Storer

To place new plans in memory, Storer needs

. Index them under the same features that the Retriever uses to find them, the goals they satisfy and

the situation in which they are appropriate.

Repairer

To repair failed plans and describe them to Storer, the Repairer Requires

1. A vocabulary of plan failures.

2. A set of repair strategies that are indexed by that vocabulary.

Storer revisited

To place new plans in memory, Storer needs

. Index them under the same features that the Retriever uses to find them

1. The goals they satisfy.

2. The situation in which they are appropriate.

3. The problems they avoid

Assigner

To decide which features in a situation are to blame for a failure, the Assigner needs

Be able to describe the cause of the failure. The more extensive its vocab the more exact its credit assignment.

Anticipator

Anticipator take the features of the situation that the planner is dealing with, and recall problems that have arisen in past situations to handle them.

Thanks!

Muhamad Heshammhesham.fcis@gmail.com