News

05 March 2010

Final results of RSCTC'2010 Discovery Challenge are published

RSCTC'2010 Discovery Challenge is over. Final results have been published and can be found on leaderboards of Basic Track and Advanced Track respectively. Official announcement will be published soon in challenge news at Overview pages.
Congratulations to the winners!
28 January 2010

History of submissions

Participants who submitted one or more solutions to a challenge may now see the full history of their submissions, including submission dates, names of solutions and their preliminary results. The history is shown on Submit page of the challenge, below submission form. We hope that having the history in one place will help participants keep track of their different approaches and come up with even better algorithms! :-)
13 January 2010

Preview of error log on Leaderboard

If you have submitted a solution to a challenge and evaluation failed for some reasons producing "error" result, you can now see the details of the error and find out where the implementation bug is hidden. For this purpose, simply click the "error" link on Leaderboard. A new window will show up with a stack trace of the exception that caused evaluation to terminate.

Note that the error log is disclosed only to the author of the solution, so you must be logged-in to have a link under "error".
01 December 2009

RSCTC'2010 Discovery Challenge is open!

RSCTC'2010 Discovery Challenge is a special event of the 7th International Conference on Rough Sets and Current Trends in Computing that will take place in Warsaw, Poland, June 28-30, 2010. The task is related to feature selection in analysis of DNA microarray data and classification of patients for the purpose of medical diagnosis and treatment. Best solutions will be awarded with $3,000 of money prizes and coverage of RSCTC'2010 registration fees.
06 November 2009

TunedTester 1.6 released

New features in TunedTester 1.6:
  • With --print-stack-trace command-line option TT displays full details of errors that occurred during test.
  • Time limit of test execution can be defined directly in GUI, not only from command line.
  • Current date and time or every test printed in the test log, to facilitate interpretation.
From now on, TunedIT server does not support TunedTester 1.0 any more, so please upgrade.
02 November 2009

RSCTC'2010 Discovery Challenge starts soon

Stay tuned for RSCTC'2010 Discovery Challenge that will be launched at TunedIT. This data mining contest, associated with rough-set conference RSCTC'2010, will stimulate data-mining research in the area of feature selection algorithms and will promote open research practices fostered by TunedIT. The task will be related to bioinformatics and analysis of DNA microarray data. Prizes will be awarded to the best solutions.
11 October 2009

TunedIT has discussion forum

We invite everyone to TunedIT bulletin board, where you can post questions and discuss all issues related to TunedIT, TunedTester, Repository, Knowledge Base, automated evaluation, reproducible experiments, data mining, machine learning, knowledge discovery, computational intelligence, ...
01 October 2009

TunedTester 1.5 released

New features in TunedTester 1.5:
  • Every single test is started as a separate process instead of a thread. It is possible now to safely stop a test in the middle of execution without risk of TT hang-up or inconsistency of runtime environment (Java threads cannot be safely terminated).
  • Memory limit for a single test can be defined using --memory-limit command-line option.
  • StandardLoader.loadData() accepts datasets that are represented by Java classes. If you have a dataset stored in an atypical format, you may put it into a JAR file accompanied by a wrapper class that reads this dataset and returns data samples in Debellor representation. Such a class can be used in place of a dataset in TunedTester test specification. For examples see the classes that read MNIST OCR data, in Swietlicka/MNIST.jar.
17 August 2009

TunedIT becomes publicly available

Steady, ready, ... GO!

TunedIT is available now for everyone. The whole world can use our cutting-edge tools to conduct reproducible data mining research and devise algorithms of the best, previously unseen, quality!
Copyright © 2008-2010 by TunedIT
Design by luksite 2010