Posted by: technofutures | November 10, 2008

Cambridge Michaelmas Week 4

One more busy week, as we pass the half-way mark in term! Only 3,5 weeks left! Lots of work on my project this week, a careers fair, an IET presentation and a lovely lunch with Hugo’s family yesterday!

Careers Fair

It’s that time of term, where the University Careers Service advertise their annual careers fair. I don’t usually go, but was intrigued this year to take a look… Lots and lots of companies this year… and the usual dodging of questions of the style ‘Are you interested in XYZ?’! The highlight, I think, was the InFusion Development stall… specialized in Microsoft Surface software development, they had even brought one to demo! Among the examples were a Tetris game developed using the recently announced Surface SDK! I’m going to try and get them to come and give a talk in the department…

IET Presentation,

Thursday evening, a last minute decision led Hugo and myself to attend a lecture by on developing a ZigBee ‘Connected Home Product’  (Zigbee is a series of protocols which provide services similar to Bluetooth over short ranges). The talk covered aspects of Home automation, but I thought the lecture was a little general. sadly very few specifics were discussed… Shame. There seems to be an interesting talk coming up in January ‘Manufacturing Flexible Displays’ by Plastic Logic.. mustn’t forget to go…


The best lecture this week? Although I really enjoyed the control systems lectures again this week (you may notice a trend…) the best lecture was probably the last innovation lecture which we had on Wednesday, which covered ‘the internet’ as an innovation case study, and concepts such as Market tipping, market types, types of adopters… The guest speaker in the second half of the lecture, John Naughton, professor of Public understanding of Tech. at the Open University, was very good and he raised interesting points about closed and open platforms (comparing the iPhone and the G1…) and he created a good discussion about the evolution of the internet.


Lots of progress this week. I think I’ve finally successfully implemented recognition algorithms based on (1) Canonical Correlations, (2) Gaussian Modelling and the Kullback-Leibler Divergence and (3) Resistor Average Distance (extension of the KLD algorithm). Below is an illustration of data that I collected, showing how well the various systems work and their relative merits…


I’m still trying to implement a system based on Maximally Probable Mutual Modes, and am getting all sorts of errors in Matlab, mostly caused (as always…) by numerical instability… I’ll figure it out… eventually… with my supervisor’s help : )

At this stage, I think I’ll be exploring a new avenue. Having explored several statistical techniques for comparison of image raw-pixel based feature vector comparisons, I’m going to try exploring other representations for the images I have e.g. representing them as a vector of ‘features’, such as those that may be obtained using SIFT (See my first post) or Histograms of Oriented Gradients… More in the weeks to come!

Till next week… or maybe not! I will be spending the whole of the next weekend working on my Michaelmas term project presentation, so it’s unlikely that I will write my blog entry!



Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s


%d bloggers like this: