AN OVERVIEW OF AGENT BASED PARADIGMS AND ENTERPRISE SIMULATION FOR ELEARNING AND KNOWLEDGE TRANSMISS - PowerPoint PPT Presentation

1 / 20
About This Presentation
Title:

AN OVERVIEW OF AGENT BASED PARADIGMS AND ENTERPRISE SIMULATION FOR ELEARNING AND KNOWLEDGE TRANSMISS

Description:

AN OVERVIEW OF AGENT BASED PARADIGMS AND ENTERPRISE SIMULATION FOR E-LEARNING ... Knowledge transfer is 'maieutical' Marco Remondino. remond_at_di.unito.it ... – PowerPoint PPT presentation

Number of Views:95
Avg rating:3.0/5.0
Slides: 21
Provided by: Marc148
Category:

less

Transcript and Presenter's Notes

Title: AN OVERVIEW OF AGENT BASED PARADIGMS AND ENTERPRISE SIMULATION FOR ELEARNING AND KNOWLEDGE TRANSMISS


1
AN OVERVIEW OF AGENT BASED PARADIGMS AND
ENTERPRISE SIMULATION FOR E-LEARNING AND
KNOWLEDGE TRANSMISSION
  • Marco Remondino
  • remond_at_di.unito.it
  • University of Turin, Italy

2
We will deal with
  • methodologies for designing models of interactive
    simulation, based on software (artificial) agents
  • to be used for education and in general for
    knowledge transmission through e-learning
    techniques

3
E-LEARNING
  • The term e-learning refers to the use of
    computers and related technologies, such as
    networks, multimedia and so on, for educational
    purposes
  • A model is built, representing some real
    situation in which the learners can take
    decisions and see the corresponding results.

4
Two Agent Based Paradigms
  • Reactive Agents
  • No internal representation of the environment
  • When to act? What action to choose?
  • Aggregate behaviour (consequence)
  • Cognitive Agents
  • Presence of high level goals
  • Behaviour as a consequence
  • Learning by trial error

5
Action Selection and RL
  • action selection problem at time t1
  • How to select an action?
  • Reinforcement Learning turning a quantitative
    value (payoff) into behavioural pattern

6
How to Select an Action?
  • Q-Learning like algorithms (expected value
    function for each action)
  • Agents wellbeing metaphor (internal drive
    external stimuli)
  • Evolutionary Algorithms (aggregate level
    natural selection)

7
Interactive Enterprise Simulation
  • An implementation of virtual tutoring by
    intelligent agents
  • Learners (tipically students) can practice the
    model
  • Knowledge transfer is maieutical

8
Management Game Vs Virtual Tutoring
  • A management game is one in which the player
    takes decisions and observes final results
  • A virtual tutoring system is similar, but with
    the addition of some entities, able to aid him
    finding the right patterns (simulated decision
    support system)

9
An Operative Example
  • http//e-lab.di.unito.it/SimulazioneAziendale
  • An advanced management game (using System
    Dynamics)
  • Now developing a framework of virtual tutoring,
    with intelligent agents (beta stage)

10
You Take the Decisions
11
You Observe the Results
12
..and the Agent can give hints and explanations
  • Hints
  • Your production capacity is too low, improve it
  • Explanations
  • Profit went down since your production capacity
    is too low

13
How to do that?
  • Sentences are actions for the agents, to be
    selected via a Q-Learning like mechanism, with
    the model parameters as input
  • Agents practice the model and find the best
    action to choose, according to the observed
    patterns

14
Another Point of View
  • Besides being used as a Virtual Tutoring System,
    intelligent agents can be part of the model
    itself
  • Social systems are complex and not always
    deterministic agents can be the building blocks
    for them

15
Data Base Security
  • A framework to explain and teach the basic
    parameters of DBS
  • We have a set of agents, which are the users of a
    database, organized into a hierarchy
  • The learner is the DB manager and can change some
    core parameters

16
The Learner can
  • Change the privileges on data, for the different
    levels of users, keeping in mind that
  • High level users have less chance to corrupt data
  • The number of low level users is higher, so they
    need to access the data more frequently

17
Data Integrity Vs Efficiency
  • The user must change the parameters in order to
    find the optimal tradeoff

18
Time for Accessing a Datum
  • Flexible (unsecure) situation
  • Read and Write privileges are always given
  • Unflexible (secure) situation
  • The permission must be asked to higher levels and
    so on. Time is

19
Conclusion
  • Intelligent Agents to be used for E-Learning and
    knowledge transmission
  • Virtual Tutoring Systems and Management Games
  • ABS frameworks to experience on simulated social
    systems

20
Thank you!
  • Thank you for your attention
  • This presentation is also a proposal for
    interaction
  • Email remond_at_di.unito.it
Write a Comment
User Comments (0)
About PowerShow.com