Monday, July 7, 2008

Some hicking pictures... and videos





Day6: American independence day celebration







Its better to see the final performance.

Day5: Beginning of the mission

I am not updating the blog everyday. I did not do the update for more than 3 days and today I am going to add all materials related to talks and happening on July 3rd 2008.

I have noticed that I am sometimes getting easily tired these days. It is mainly because I am not sleeping at the appropriate time. I need to correct on this issue.

Coming back to today. It was a day filled with talks and work-group meetings. I tried to attend as many workgroup to get a glimpses of various possible projects and collaborative work. But overall it was a fruitful day with many interesting talks and ideas.

First person to open the day was Lloyd Watts, CTO of Audience. It is interesting to note that he was also somehow associated with Carver Mead from Caltech. Another interesting thing is that he is also one of the few person working hard to commercialize Neuromorphic Engineering Technology. Its been a long complaint from my heart that this area did not have any company that demonstrates the practicality of neuromorphic engineering technology. Without this authenticity is lost and the effort becomes more scientific rather than engineering exercise. He gave a good talk and demonstration about how his idea about Cochlea model evolve into a product. I was really impressed by his talk. One intereseting point that he raised later was that if a startup need to succeed than its mainly because of the good rapport and relation that the startup makes with the big company which will ultimately sell the product. This rapport is important and efforts need to be taken by means of sincerity, commitment, helping the customer companies realize the goal. Along with excellent technology all these things are need for success of a startup firm.


The next talk was given by Paul Hasler from GA Tech. It was an interesting talk about the applicability and validity of using Analog Arrays called FPAA for simulating large circuits. He was quite good in explaining the principles of FPAA and its evolution. As such many of the ideas presented in the work is useful even if we go for an application specific architecture for neuromorphic systems.

The third talk of the day was given by Chuck Higgins from Univ. of Arizona. It was mainly about his research on understanding the brain of various insects and modeling them. It was interesting to know the kind of effort that he took to design system for easily tracking and managing behaviour of insects. I dont remember much of the talk and would recommend you to look at the slides ( i will post all the presentation materials soon)



The remaining part of the day focussed on various issues and technology.
First a small demonstration about using Subversion System called Tortuise SVN.
Next a brief introduction about USB and associated tools for USB drivers/firmware.
Then we had a brief chat about spike based processing and computation challenges.
Following that we had the first session on GP-GPU.

It was day filled with excitement and ideas and inspiration.

Hope in the remaining days we accomplish something worthwhile.

Thursday, July 3, 2008

Day4: A day of ups and downs

Today it was filled with many presentation but only a few of them was of interest to me and hence I did not give much attention to other talks.

I was quite impressed by the continuation of the talk given by Prof. David Heeger from NYU. He had some interesting insights into the principles of attention mechanism behind vision. He continued to talk about Neurocinematics (using fMRI to analyze the human response for different kind of movies..it was fun and at the same time revealed some good insights into many aspects of vision and temporal dimension in vision)

Further talk was given on Neuroelectrical stimulation and its aid in recovery from Spinal cord injury. It explained many principles of Spinal Cord and the various nerve mechanism in Spinal Cord. How to analyze, process and encode the electrical signal coming from Spinal Cord nerve fibers. The talk was given by Prof. Jimmy Abbas from ASU.

Two of the remaining talk was on Olfactory signal processing in Bees and on some other topic. It was totally unrelated to me and so I was not much interested in it.

I also spent more than 6 hours struggling to get CUDA working on GNU/Linux Ubuntoo 7.10. It was a big pain in the ass with so many problems that I cannot describe. Somehow it worked in the end. Here is the configuration that finally worked. All of these are important.

OS: Ubuntu 7.10
Gcc: 4.2
CUDA toolkit and SDK: 2.0 beta version
CUDA driver: 171.xx ( 173.x did not work. 169.x was too old)
Machine had 2 Tesla card and one simple Nvidia card.
A detailed instruction on what to do after downloading the driver from nvidia and to uninstall existing driver in ubuntu is given in

http://ubuntuforums.org/showthread.php?t=685196
http://cutecomputer.wordpress.com/2007/06/05/installing-nvidia-graphics-card-on-ubuntu/
http://lifeofaprogrammergeek.blogspot.com/2008/05/cuda-development-in-ubuntu.html

Wednesday, July 2, 2008

Day3: Second day of the workshop

Another day went past so quickly.

Now things are moving much faster than I experienced in the first two days.

I was getting familiar with many people in the workshop and we were starting to enjoy the company of each other. Also today's talk was fairly okay. Out of the four lectures that was presented today, one was very good, one was good, one was poor and I skipped the last one (I skipped part two of the lecture on locomotion by Cohen).

The lecture that was very interesting and useful covering many details of the visual processing was given by Prof. David Heeger from New York University. He tried to cover many insights on the visual trajectory especially from the Cortex point of view. Since I have read the Eye. Brain and Vision by Hubel, I found the lecture easy to follow and understand. He pointed out some good insights into the higher levels of the brain from MT onwards. I would give him 4/5 for his excellent presentation, interesting outlook and for his nice Q&A sessions.

The other talk in the morning was given Lick Lyon, a student of Carver Mead from Caltech (In fact majority of the speakers were direct descendant's of Mead). He went into the details of his analytical modeling of Human Cochlea. His talk was interesting and had many historical perspective and touched mainly on quantitative evaluation of Human Cochlea from filter bank perspective.

The remaining two talks was given by Avis Cohen and Tara Hamilton. Tara Hamilton (pics below) had some interesting ideas but overall I did not like the talk because of lack of focus or depth or clarity in the presentation.
Later in the evening a avalanche of topics related to Neuromorphic Engineering was presented. I will mostly focus on the spike based processing scheme and its related topics. I was also thinking of giving a small talk on GPGPU computing and using GPGPU for simulating neuromorphic principles.

One interesting visitor for this workshop was a small girl ( I need to find her name).
In the night we had a small get together in our house and one of my room mates (Max) prepared the food for the night. It is good that he initiated the idea as the home became more lively and we got to know each other in the house well. Thanks to Max for his initiative and his effort.

(Above picture from the windows of our lecture room)