Soft Computing - A Research Perspective

Recently,I attended this 2 day conference on SOFT COMPUTING - A RESEARCH PERSPECTIVE.

Venue: Dept. of Information Science & Engg, REVA Institute of Tech and Mgmnt, Bangaore.

Date: August 13 / 14, 2010.

Highlights of the Conferences / Workshop:


Day - 1:
Keynote Address: Y. Narahari, Head - Dept of CSA, IISc, Bangalore.

Highlights of his session:
Game Theoretic Approaches to Knowledge Discovery and Data Mining (KDD).

• Recent Interest in Game Theory based techniques for solving KDD problems.
• Some relevant notions in Game Theory and implications for KDD problems.
• Example of KDD problems
• John Von Neumann
• Applications of Game Theory to KDD problems
• Machine Learning - Social Network Analysis and Social Network Monetization.
• Game Theory models are very natural for many KDD problems.
• Nash Equilibrium
• Shapley Value
• Bounded Rationality
• Open Source Tools - Gambit

Gambit: Software Tools for Game Theory

Gambit is a library of game theory software and tools for the construction and analysis of finite extensive and strategic games. Gambit is designed to be portable across platforms, and runs on Linux, Mac OS X, and Windows.

Gambit provides:

• A graphical user interface, based upon the wxWidgets library, providing a common interface with native look-and-feel across platforms.
• A library of C++ source code for representing games, suitable for use in other applications.
• A Python API for scripting applications.

Read More on Gambit - Game Theory

Comparison with Gambit and other Tools. NECTAR is implemented in Java, which provides. platform independence. Read on...

Session - 2

Introduction to Machine Learning

Resource Person: Indrajit Bhattacharya, Asst Prof, Dept of IIsc, Bangalore.

Machine Learning - Lean from Experience

Highlights:

• What is Machine Learning?
• Why do Machines need to learn?
• Human Expertise may not exist
• Humans unable to explain expertise
• Insight or knowledge may be hidden in vast amounts of data.
• Roots of Machine Learning
• Statistics and Probability - Decision Theory, Estimation Theory
• Brain Theory - Neural Networks
• Psychological Model
• Evolutionary Model - Genetic Algorithms, Artificial Intelligence
• Adaptive Control Theory - Reinforcement Learning

Machine Learning in Action:

• Spam filtering
• Fraud Detection 
• Speech and Handwriting Recognition
• Mars Rover
• Deep Blue
• Customer Profiling
• Automated Driving

Applications of Machine Learning:

• What is to be learnt?
• Functions / Distributions
• Rule Sets

Learning Styles:

• Supervised Learning
• Unsupervised Learning - Clustering
• Semi - Supervised Learning
• Re-Inforcement Learning

Learning Approaches 

• Neural network / Kernel Network
• Linear Model
• Decision Tree

Tools:

• Weka - Weka is a collection of machine learning algorithms for solving real-world data mining problems. It is written in Java and runs on almost any platform. The algorithms can either be applied directly to a dataset or called from your own Java code.
• Rapid Miner

Session - 3:

Role of Animals in solving Engineering Problems

Resource Person: S N Omkar, Dept of Aerospace, IISc.

Highlights:

• Swarming is powerful
• Artificial Bee Colony
• Ant Colony Optimization
• CACS - Continuous Ant Colony System
• PSO - Particle Swarm Optimization

Day - 2

Computational Intelligence in Mobile Wireless Networks

Dr.Vijay Kumar B P, Prof & Head, Dept of CSE, REVA ITM, Blore.

If you want to be incrementally better, be competitive 

If you want to be exponentially better, be co-operative  - Anon

Highlights:

There are mainly 3 major categories of problem computation involved in this world:

1. Classification
2. Functional Approach
3. Optimization

Hard vs Soft Computing

Soft Computing:

Tolerant to imprecision, uncertainty, partial truth and approximation. 

Ex: Human Mind

Hard Computing:

Conventional Computing, requires a precisely stated analytical model.

But, the real world problems exists in non-ideal environment.

Soft Computing Premises:

• Fuzzy System
• Neural Networks
• Genetic Algorithms
• Swarm Computing 
• Machine Learning
• Probabilistic Reasoning

The real World problems are pervasively imprecise and uncertain.

Unique property of Soft Computing: "Learning from experimental data"

We are drowning in information and starving for KNOWLEDGE - Rutherford R.

Session - 2:

Support Vector Machines

Resource Person: Shirish Shevde, Dept CSA, IISc, Blore.

PDF Resources of the session

Session - 3

Neural Computation

Session by :Prof. T M Nagabhushan, Special Officer, E - Learning, VTU.