I found this link by Matt Gershoff highlighting some really good link on ML. This includes topics like LDA, lectures by MJ on frequentist vs bayesian, deep learning.
The author has also highlighted that not knowing linear algebra while doing ML is like not knowing where the milk comes from while making cheese. So he has stressed it’s importance. I found in his link section someone posting about a short book he has written on linear algebra and it looks to be good
Youtube has really interesting series of documentary on cold war starting from 1918-end. If you are interested it would be a good link to visit and know more about world history.
The author has given a very neat review of different process for doing non-parametric bayesian clustering. I am reading on this for the first time and the take home message is that these models have the advantage of increasing the number of clusters as we get more data-points. Do have a look the following clusters discovered by the algorithm from the list of food items (features as nutritional content) from Mc-Donalds.
Feels good to write a post after a long time. I am having a good time india, exploring, eating and snoring.
I have to submit camera ready version of my FG2013 paper by Jan’15. Since we had a bunch of videos as supplementary material with the paper, Abhinav suggested me to put these videos on my youtube channel and share the link. This will be better than hosting videos on my website since it might be down sometime or become redundant later. I realized the efficacy of this idea once I created a playlist and put these videos for there. The playlist is as follows (wait for all of them to run). The description will be available after you open the link in youtube
I have been in a state of confusion. This phase often comes after the completion of one project and thinking about next project or direction. I cannot generalize this to all researchers but I often get plagued by this. I know of researchers keeping themselves busy with multiple projects, which might prevent them from `this’ bewilderment.
This phase starts when you begin to rationalize that your last work was much below your expectation (even if it is good!) and slowly begin to pursue a bigger picture for your research plans. Other symptoms are that I typically start thinking about challenging myself with a difficult problem and often try to devise ways to learn something that I don’t previously know. When I say ‘devising ways’ it means that I might try to look for a problem that requires learning a method that I previously don’t know and learning this method might require more effort than usual. It might seem like a fun period from description but it is not since there is a lack of focus, less work and the guilt of not doing something concrete hits you repeatedly. You might also think of this as a time when you start thinking of a new idea or next step for your thesis. Although I might seem to be cynical about this phase but on the positive side it gives me a chance to realize that I might be maturing as a researcher by seeking a broader perspective in my research and I also get a chance to read more than usual. The advantage of reading more is that after a time your mind starts to arrange and connect things with each other which I often refer to as `deconstruction’.
Referring to this as the confusion phase, I don’t have much intuition on what is the best way to proceed and re-building your focus. For me this phase has prolonged a bit this time owing to a departmental examination (prelims) and an incoming trip to India. May be in near future as I get more data points and experience I would be able to shed more light on this. Till then good luck.
I would like to thank some of my friends for liking this blog and encouraging me to continue. Whosoever it is in the world (except a few) needs some motivation and I am one of them.
I was trying to listen to classical Indian fusion on YouTube today and found a certain limitation during this process. Compared to Pandora (which is music recommendation service like lastfm) listening to songs on YouTube is limited. I don’t have the option of starting my own station and relax and listen.
The prime difference between YouTube and Pandora would be the limited number/expressiveness of tags that YouTube videos have or which are being currently used. For instance a classical song on YouTube is often tagged as just ‘music’ or ‘entertainment’ w/o any regards to it’s musical or video content. On the other hand Pandora has pretty descriptive tags for it’s songs, which makes it easy to find songs in same genre even though these songs might not have the same artist or usual features like name-viewers. Moreover during a small glance I also observed that the options/suggestions that YouTube provides, while viewing a video, have a similar name/viewers, than content based features like classical, rock, or other.
Thus in future it will be interesting to see if Google could use these noisy tags in an intelligent way so that a user can use YouTube depending on his preference. Thus one one day I can use it as Pandora that recommends music based on similar content or like the current YouTube. Google has also been doing some related work here, one of which I came through was published in ICCV’11 by Thomas Leung and others on using weakly supervised learning for handling noisy labels/tags (noise is other source of nuisance). Another homework that I should do is read about the recommendation system that Google used and if they are doing something similar to those employed by online retail sites that use algorithms like latent models, matrix factorization etc (mostly unsupervised). I will revisit this discussion in future.