A system analyst by training, I have more then 14 years of teaching and software development experience in open source as well as proprietary technologies. An avid pursuer of knowledge, I have gone on to complete my Master studies in 2013.

A prolific writer and solution provider to codeproject.

2014 -
* Best Mobile Articles of September 2014 - Second Prize

* Best Mobile Articles of August 2014 - Second Prize

* Second prize winner in Article Competition on Android development organized by the codeproject.

* Winner in Article Competition on HTML5 and CSS3 organized by the codeproject and garnered 4 winning and 2 honorable mentioned articles.


Web Application Developer


Finding Global Minimum in Rosenbrock's Valley on MATLAB

The Rosenbrock function, introduced by Howard H. Rosenbrock in 1960, is a non-convex function used for testing optimization algorithms. It is also known as Rosenbrock's valley or Rosenbrock's banana function. It is defined as:
f(x,y)=〖(1-x)〗^2+100〖x(y-x^2 )〗^2

The global minimum is inside a long, narrow, parabolic flat valley
where f(1,1)=0.

This video demonstrates the finding of the global minimum via gradient descent method which is a very slow but effective process.

Q-Learning for World Grid Navigation on MATLAB

Q-learning is a kind of Reinforcement Learning that assigns values to action-state pairs. In every state there are a number of possible actions that could be taken and each action within each state has a value according to how much or little rewards the robot will get for completing that action.

In this exercise, a robot traverses on a 10x10 grid from the start state on the top-left cell to the goal state on the bottom-right cell using Q-learning algorithm.

Without a teacher, the robot will move from state to state through series of actions involving exploration and exploitation until it reaches the goal. Each cycle from start state to end state constitutes a trial. Each transition between states will see the robot moving up, right, down, or left. Whenever the robot arrives at the goal state, the program resets the robot to the start state to begin the next trial while retaining the memory of its prior learnt knowledge in the form of Q-values.

At trial one, the robot takes off without any prior knowledge of the worth of taking any actions, i.e. the Q-values are all set to be 0. These Q-values will continue to be updated from state to state and from trial to trial until it reaches a convergence stage where there is no marked improvement in Q-values between two consecutive trials.

Qualifications & Certifications

Master of Technology in Knowledge Engineering

National University of Singapore

Graduate Diploma in Systems Analysis

National University of Singapore

National University of Singapore

National University of Singapore

Skillpages has been acquired by! is pioneering the way people find local services. Skillpages is the world’s premier directory of service providers.

Find out more

Peter Leow provides the following services:

Supported Countries
Choose your country