Harris Georgiou - Homepage
fMRI-Sparse toolbox for Matlab PDF Print E-mail
Written by Harris Georgiou   
Saturday, 22 June 2013 00:00

The ASSURANCE project is about developing new state-of-the-art algorithms for sparsity-aware distributed models, mainly for Machine Learning in fMRI data decomposition/analysis and in telecommunications.

ASSURRANCE project is about developing new state-of-the-art algorithms for sparsity-aware distributed models, mainly for Machine Learning in fMRI data decomposition/analysis and in telecommunications.

There are several currently running code and data development side-projects linked with the main ASSURANCE project. These are some of the resources I have personally developed and can publish here as-is (read the licensing info included).

fMRI-Sparse toolbox is a minimalistic collection of low-level data handling (matrix) functions for fMRI processing, block-based & event-based test pattern series, as well as "realistic" simulated fMRI data series for algorithm benchmarking, template scripts for various fMRI decomposition methods (GLM, PCA, ICA, BP, KSVD, ...), analysis of components & activation maps, etc. Since the toolbox can be used as a benchmarking suite, several data generators are included for creating fully-identifiable fMRI-like data series.

The main implementation platform is Matlab, currently using toolboxes from versions 8.0+ (releases R2012b and above), e.g. the Basis Pursuit algorithms from Wavelet Toolbox. However, with some small conversions in the source codes, the main functions should be fully runnable with other Matlab-compatible platforms like Octave (GNU). Also, there are some external codes and toolboxes required to run some methods (e.g. fastICA), which too can be substituted by other similar packages if necessary.

Please note that these resources are under continuous development and updates (beta versions), so some features may not yet be fully implemented or stable.

READ MORE >>

Last Updated on Tuesday, 24 September 2013 19:40
 
Security Alert: Emergency turn-off of web Java now mandatory PDF Print E-mail
Written by Harris Georgiou   
Monday, 28 January 2013 00:00

Inormally do not put emergency security alerts in this pages, but due to the severity of this event, I thought that everyone should be warned as soon as possible:

"Oracle releases emergency Java patch; experts warn flaws may take 2 years to fix" (networkworld.com)

Security Alert: Emergency turn-off of web Java now mandatory

Unfortunately, it seems that the worst fears about Java’s severe security holes have now become true.

Since the exploits are already included in various publicly available exploit kits (e.g. “Blackhole”) and the problems can not be fully addressed for the next 12-24 months, the best options right now are, from most drastic (and secure) to the mildest (and more dangerous):

  • uninstall Java completely, if not required on any Internet-connected device
  • disable web Java (JRE) in the browsers, from the Java control panel
  • leave web Java enabled, but disable the related plug-ins in the browsers
  • leave web Java enabled, as well as the plug-ins, but review every prompt carefully

The last option relies only on the fact that the latest patch (7u11) from Oracle sets the default security level to "high", so every Java applet will trigger a dialog prompt for the user before it is executed. Keep in mind, though, that this is the most dangerous option, since it takes one single successful attack to breach local security and enable full remote access to the device (not just infect it with some virus or spyware).

Status update (1-Feb-2013): "Oracle Responds to Java Security Flaws with 50 Fixes" (new version: "7u13")

Status update (2-Mar-2013): "New Java 0-Day Vulnerability Being Exploited In the Wild" (latest versions: "6u41/7u15")

Last Updated on Sunday, 24 March 2013 17:10
 
Skynet is not here yet PDF Print E-mail
Written by Harris Georgiou   
Wednesday, 27 April 2011 00:00

Well, the date has come and passed without any "incident", of any short, from any renegade AI war machine - at least not something so big that would hit the media all around the world. In case you missed it, the 21st of April was the critical date, the "Judgment Day", according to James Cameron's blockbuster movie series "Terminator".

The Terminator (T2) endoskeleton...

The 19-21 April 2011 dates come from the alternate timeline of the corresponding TV series (2008), but the overall concept is of course the same. As the target time gets closer and soon passes by, without any major breakthrough in AI and, of course, no machine-initiated nuclear war, the screenwriters have to readjust their expectations and prolong our fears. Since the original story is already three decades old, the latest movie of the series adopts the post-attack viewpoint, showing that we are indeed at war with the machines, as always.

Despite all that, Cameron himself seemed preoccupied with other things right now, in a message he posted in his Twitter account that night:
"Skynet was supposed to go operational tonight. Instead of machines taking over, we have the very real threat of global warming."

READ MORE >>


NOTE: The site is currently in testing (beta) version and several pages are still under heavy construction or empty. Please stay tunned for frequent updates and of course feel free to send feedback on any bugs, broken links, etc.

Last Updated on Friday, 02 September 2011 11:14
 
Underwater Wireless Communications PDF Print E-mail
Written by Harris Georgiou   
Wednesday, 04 January 2012 00:00

Underwater communication has been a major technological issue since the start of autonomous (scuba) divers and deep-sea vehicles (DSVs). In recent years, wide-area sensor networks have enabled the close monitoring of environmental and extreme weather conditions, as well as emergency alerts in case of tsunamis and volcanic ash clouds. However, these networks usually require long-range communications and expensive satellite uplinks, in order to send data to a base station, since a fully-wired grid of hundreds of sensors over areas of thousands of square km is not a feasible solution.

Communicating underwater by hand signals

Sea water has very different signal propagation properties than open air. The major problem with electro-magnetic (RF) signals under water is that they fade very quickly with distance. A typical WiFi hotspot of some 100-200 mW transmission power becomes 10-20 times weaker, which practically makes it no more than a Bluetooth-like ultra-shortrange headset.

These technological challenges have boosted research interest during the past decade or so. An interesting introductory text regarding these issues is one by D.Pompili and I.Akyildiz, "Overview of networking protocols for underwater wireless communications" (IEEE Communications Magazine, Feb.2009, vol.47, n.1, pp.97-102 - doi:10.1109/MCOM.2009.4752684), full text (pdf) available from Rutgers, The State Univ. of New Jersey. There is also a similar paper by Mari C. Domingo, "Overview of channel models for underwater wireless communication networks" (Physical Communication, vol.1, n.3, pp.163-182, Sept.2008- doi:10.1016/j.phycom.2008.09.001)(needs access to ScienceDirect.com).

Last Updated on Wednesday, 04 January 2012 19:33
 
Scientific blogging: Is it worth it? PDF Print E-mail
Friday, 17 September 2010 00:00

Web 2.0 technologies have enabled the merging of different types of publications into one, the Internet, with millions of authors and billions of readers. Science is not an exception. Countless articles are presented in the Web as “scientific” in content, but there is a thin line between solid science and speculation. Are personal blogs appropriate for such everyday-science content and, more importantly, can they be trusted as true?

Well, blogs are not meant to publish research papers. Everybody knows that. The reason is not because of the content or the medium; it’s because of the intended audience. Many respectable public repositories of scientific papers like arxiv.org and citeseer.net contain a vast amount of true scientific knowledge. Scientific publications like Nature, Scientific American and New Scientist have their own blogging sections. Dedicated portals like Science20.com and Scientceblog.com publish thousands of science-related posts daily. There are even Facebook-like social networks for scientists, like ResearchGate and Academia.com.

There is a notorious “guide for authors” (pdf) for those who wish to actually publish a research paper in one of IEEE’s scientific journals. It illustrates, in a very graphic and hilarious way, the difference between a simple statement like “1+1=2” and the way it should be presented in a more impressive scientific form. The sad thing is, it’s not very far from the truth! Everyone who’s working on research and writes such papers has at least a few similar stories to tell, where a submitted paper was rejected in one journal as insufficient, only to get published in another with honors.

How to become an IEEE author (1+1=2)

Do you think the “IEEE guide for authors” above is correct? Think again. The running variable z in the limit in the definition of e is incorrect, it should go to infinity instead of zero. If it was really a scientific paper and not a joke, this would be bad, really bad.

READ MORE >>

 

NOTE: The site is currently in testing (beta) version and several pages are still under heavy construction or empty. Please stay tunned for frequent updates and of course feel free to send feedback on any bugs, broken links, etc.

Last Updated on Monday, 27 September 2010 12:09
 
<< Start < Prev 1 2 3 Next > End >>

Page 1 of 3