Activities – Page 2 – Carlo Curino

Neural Network: implementing Lamstar

In this post i report a simple implementation of a Counter Propagation Neural Network.

As for the following examples i will post the goal is character recognition on a grid of pixels.

Noise is introduce to evaluate robustness of the approach.

Here we will present both implementation and performance eval-

uation of Lamstar neural network applied to characters recognition.

This documentation provide full description of the approach and the results.

This is the source code for the neural network

Neural Network: implementing Counter Propagation

In this post i report a simple implementation of a Counter Propagation Neural Network.

As for the following examples i will post the goal is character recognition on a grid of pixels.

Noise is introduce to evaluate robustness of the approach.

Here we will present both implementation and performance evalu-

ation of an artiﬁcial neural network based on the Counter Propagation

applied to characters recognition.

This documentation provide full description of the approach and the results.

This is the source code for the neural network

Neural Network: implementing Hopfield NN

In this post i report a simple implementation of a Hopfield Neural Network.

As for the following examples i will post the goal is character recognition on a grid of pixels.

Noise is introduce to evaluate robustness of the approach.

Here we will present both implementation and performance eval-

uation of an artiﬁcial neural network, namely the Hopﬁeld network

applied to characters recognition.

This documentation provide full description of the approach and the results.

This is the source code for the neural network

Timeplot: an MIT tool to visualize series of events

Timeplot is a DHTML-based AJAXy widget for plotting time series and overlay time-based events over them (with the same data formats that Timeline supports). The original site is: http://simile.mit.edu/timeplot/

I’ve installed it and make a couple of very simple tests, the installation is very simple and the XML-based language to feed the graphical interface very intuitive here is an example.

Here follows a simple example of XML input file:

<?xml version="1.0" encoding="UTF-8"?>

<data>

<event start="Jan 01 1849 00:00:00 GMT" title="California Gold Rush" link="http://en.wikipedia.org/wiki/California_Gold_Rush"></event>

<event start="Apr 12 1861 00:00:00 GMT" end="Apr 09 1865 00:00:00 GMT" title="American Civil War" link="http://en.wikipedia.org/wiki/American_Civil_War"></event>

</data>

Neural Network: implementing backpropagation

In this post i report a simple implementation of a Backpropagation Neural Network.

As for the following examples i will post the goal is character recognition on a grid of pixels.

Noise is introduce to evaluate robustness of the approach.

Here we will present both implementation and performance eval-

uation of a multilayer artiﬁcial neural network, exploiting backpropagation for characters recognition.

This documentation provide full description of the approach and the results.

This is the source code for the neural network

Neural Network: implementing Madaline

In this post i report a simple implementation of a Madaline Neural Network.

As for the following examples i will post the goal is character recognition on a grid of pixels.

Noise is introduce to evaluate robustness of the approach.

Here we will present both implementation and performance eval-

uation of a multilayer artiﬁcial neural network, based the Adaline

neurons (Madaline) for characters recognition.

This documentation provide full description of the approach and the results.

This is the source code for the neural network

AI simple programs: MiniMax for game playing

Another interesting topic in AI is game playing, the MiniMax algorithm is been presented,

together with the optimized version known as Alpha-Beta pruning. Here i show an

implementation and result evaluation for both algorithm on a simpliﬁed version of the

well known Othello Game (also known as Reversi or Attaxx).

This documentation provide a complete description of the steps and

here you can find the source code for the optimized version of the program.

AI simple programs: heuristic search

Informed search algorithm are usefull when dealing with large state

space, and when ”good” heuristics can be ﬁnd for the problem. Here we

present two simple an implementation of the well known algorithms: A*.

To show this two simple game, the 8 puzzle game and the dating game has

been used, in the previous we tested 4 diﬀerent heuristics while the second

has been used to compare simple depth ﬁrst, A* and IDA*.

This documentation provide a complete description of the steps and

here you can find the source code for the optimized version of the program.

AI simple programs: searching

Searching algorithm are commonly exploited in the Artiﬁcial Intelligence ﬁeld,

here I present two simple implementations of the well known

algorithms: depth ﬁrst, breadth ﬁrst and depth limited for tree search. To

show this two simple game, the 8 puzzle game and a simple grid world

has been used.

This documentation provide a complete description of the steps and

here you can find the source code for the optimized version of the program.

DrugSearch: Neural Network Information Retrieval

In this post is summarize the work i’ve done as Final Project for CS580 Neural Network in Chicago for Professor Graupe. This work has been later extended together with professor Clement Yu and published at CIKM 2005. General purpose search engines have provided the users with simple way of querying massive sources of information eﬃciently, but for domain speciﬁc problems better solutions can be developed. We present an application of the Lamstar Neural Network for domain speciﬁc web searches. The problem itself is really general and is about relating diﬀerent topics in a domain speciﬁc web search, we present here an instance of the problem relating drug names and side eﬀects. The system, working on top of existing search engines, will ﬁlter the results to improve the quality of the search via a properly trained Lamstar neural network, providing a list of links to the suitable pages while directly showing the most relevant portion of each page to the user.

Here is a presentation of the work: drugsearch.pdf