Another year, another successful bonfire night. Even Griff turned up this yer despite his fears of injury, but a good time was had by all. Siobhan got to show off her new lap-top, Carl, Cat & Sam came up to see us, and I caught flu.

During my week of illness, I made some decisions about what to do with my research over the next few weeks. Thus my "TO DO" list now includes, Finish APRON2 compiler, Research the maths behind my Parameteric models, Implement the lot on GPU technology, START WRITING THESIS!!!

You may have notices the webcam is a bit dormant. Sorry about this, but it became part of the duck. I'm waiting on an order that contains new webcams.

Hey everyone!

Wow, it was June when I last updated this blog. Truth is, I've been very busy.

In July, I was in Spain, attending the CNNA conference in Santiago De Compostella. I would like to thank Natalia for her kind hospitality, and the Judges at the conference who awarded me a "Best Student Paper" award for "A Cellular Processor Array Simulation and Hardware Prototyping Tool" (APRON) paper I submitted. You can download the paper on my research wiki. After this I went to La Coruna, to see my other spanish friend Carmen, who was undertaking her PhD final examination. She got it, so congratulations to her - Dr. Carmen!

Arriving back in England I spent three weeks dedicated to the REVERB workshop, which was hosted in my laboratory. The aim of this workshop was to attempt to collate the research efforts of Sheffield, Manchester and Aberyswyth Universities, and finally nail down some specifications. APRON was used as the underlying simulation engine for the "DUCK" robot. This robot encapsulated the basic outcomes of the REVERB research project, in that an agent actively searched for food, by looking intelligently at where food should be. APRON controlled the Pan&Tilt, image sensor and implemented the DUCKs brain!. I use the word DUCK broadly, as it was the Sheffieldians who insited it was a duck. Anyway, our duck had a laser beam for a mouth, and had a single eye at the end of its arm. Make up your own mind.

After the workshop, it was time to finally complete the Online Automated Marking System for the VHDL course at the university. This web-server based tool marks schematics submitted by the student and provides them with instant feedback, and also keeps a record of the students activities and submission progress. So this year (if it all works), I won't have to spend a week hand marking 160 studen courseworks! This activity took two weeks with the help of Dr. Carmen.

So at last, I get some time to get on with some PhD stuff. I've been implementing LISSOM again in APRON, but this time trying out various parameters and connectivity structures, in an attempt to understand the inner-workings of the model. HINT: STUDY THIS MODEL WITH RETINATOPY DISABLED! It all starts to make sense. Over the next few weeks I begin work on Adaptive Meshes and a Topographic Processor Chip. I'm supposed to have finished my thesis by now, but being the lab "odd-job" man does interfere a little!

Also, I've become an advocate of vmWare - Never bother with multi-boot systems again! I've now virtualised all my unix variants and they run fast...

Finally, on a sadder note, I would like to pass my best wishes to my fiancee Siobhan's family, as they struggle through difficult times.

As part of a university event, I took part in a Myers-Brigs Psychological Test. Apparently I'm an INTP person. According to the collective knowledge of unfacts that is Wikipedia, This means:

INTP types are quiet, thoughtful, analytical individuals who don't mind spending long periods of time on their own, working through problems and forming solutions. They are very curious about systems and how things work, and are frequently found in careers such as science, architecture and law. INTPs tend to be less at ease in social situations and the "caring professions," although they enjoy the company of those who share their interests. They also tend to be impatient with the bureaucracy, rigid hierarchies, and politics prevalent in many professions, preferring to work informally with others as equals.

INTPs organize their understanding of any topic by articulating principles, and they are especially drawn to theoretical constructs. Having articulated these principles for themselves, they can demonstrate remarkable skill in explaining complex ideas to others in simple terms, especially in writing. On the other hand, their ability to grasp complexity may also lead them to provide overly detailed explanations of "simple" ideas, and listeners may judge that the INTP makes things more difficult than they are. This to the INTP, however, is incomprehensible: They are merely presenting all of the information.

INTPs' extraverted intuition often gives them a quick wit, especially with language, and they can defuse the tension in gatherings by comical observations and references. They can be charming, even in their quiet reserve, and are sometimes surprised by the high esteem in which their friends and colleagues hold them.

Not bad...

Well, for the time being I've given up coding in software. I'm sick to death of it. Instead, this week I'm coding in hardware (or VHDL to be precise). I'm working on the implementation of a basal ganglia chip using my Xilinx Virtex5 FPGA's. If I've done my sums right, the projected performance will allow the update of a 128x128 channel Leaky Integrator neuron basal ganglia with a degree of non-plastic diffusive connectivity in around 1ms. This is by far enough for use in a real-time system.

After that, I'm going to be looking at the feasibility of swapping in and out the states of the basal ganglia layers, allowing you to time multiplex different basal ganglia loops.

Finally, this feature will be integrated into APRON.

Hey, sorry about the lack of updates last month, its been a busy time, not helped by the fact my PC power supply exploded. This has cascaded into many other problems, not least the live feed being stuck!

The start of April involved a Monopoly Pub Crawl around London. My friend Hal arranged a 22-pub crawl in 11 hours, which was a mammoth task, and very good fun. Being a little hungover when the clocks change to BST and running across London to get a train back to Manchester is not that much fun. As soon as I got back into Manchester, I was off to Sheffield university for the biannual REVERB meeting. When I returned I did some further experiments with LISSOM.

I have just got back from an 11 day intensive workshop in Sardinia, Italy. Cognitive Neuromorphic Engineering (CNE), at the Cappo Caccia hotel. Lots of hard work, but some good discussion took place.

The CNNA deadline has passed. (fingers crossed...)

REVERB: http://www.abrg.group.shef.ac.uk/projects/reverb/public/
CNE: http://www.ini.uzh.ch/~giacomo/sardinia08/
CNNA: http://www.gva.dec.usc.es/cnna2008/

Hello. As part of my research, I'm implementing various topographical biologically plausible neural networks. A fun and interesting model is the RF-LISSOM self-organising orientation model, that develops from initially random connectivity, into areas of sensitivity for a given training stimuli. Traditionally, the network can be trained with all sorts of inputs, but oriented Gaussian bars give nice results. Anyway, there's loads of studff on the net about the lissom model, and if you're interested here is a good starting point - http://homepages.inf.ed.ac.uk/jbednar/spinning_small.html. In the mean-time have a look at my first crack at an orientation map for a 96x96 array, using APRON software. Ultimately, my neurons don't reduce in sensitivity over time, so it still looks a bit unorganised.



So lets move on to the next model...

Even though the version of APRON software to download has not changed for a while, it's being considerably further developed at the moment. All the linkmap related stuff has been upgraded and is much, much faster. Enhancements to the GUI have taken place, including real-time IO components. Over 10 additional functions have been added including continuous Hebbian-like learning and, perhaps controversially, the colour blue has been abandoned in favour for black!

Version 0.5 is going to be faster, more flexible and secure than ever before.

For some reason, when i searched for the CNNA'08 website, my site came first in google, so to help my colleagues who are interested in Cellular Neural Networks and their Applications, in the year 2008, I'll put a link here:

http://www.gva.dec.usc.es/cnna2008/

Hey guys. Those of you close to me will know I've been working out each morning Since January 3rd. Well 90% of mornings, depending on the preceeding nights frivolities. Given that I hate physical activity (except hike-y stuff) I've been rather suprised at just how enjoyable going to a gym in the morning can be. I've done the whole thing properly. I got a trainer to tell me what I need to work on, and he gave me quite a workout list to do. But by far the best bit is enjoying a jacuzzi and then time in the steam-room before going to work.

Well Hello! New York was just such a fantastic place, I simply could not wait to get back there, thus tickets were bought immediately. The time has come for the sequel, introducing a new character - Siobhan! We fly out tomorrow at 10am, to spend a week in the snowy movie-set city. This time, I apparently have complimentary wifi at the hotel, and so I should be able to keep some sort of blog record. If anyone I know wants any cheap stuff, text me, and I'll see what I can do.