Lend od ist

Finally I have come out of the introductory chapters forest. In mission 3.15, MacKay summons us to slay the mythical p-value. The weapons we carry are simple, but sharp and the insights look very promising. This mission will lead us to confront the p-value and its meaning when we are comparing two different models. In this post we are going to carefully go through the exercise solution and insights. First we are going to recalculate the statistician conclusion and then move on to MacKay's approach of comparing two models by how likely is the data given each of them. In a way, this post is also intended as containing part of the essence of chapters 1 and 3. Before reading on I suggest you work on the exercise yourself (for 15 minutes, even if you can't solve it): ‘If the coin were unbiased the chance of getting a result as extreme as that would be less than 7%’ We interpret this sentence as meaning that 7% is the probability of a fair coin getting 140 or mo...

That was a long list of exercises! The most interesting was surely Buffon's needle and noodle, which I couldn't solve in the proposed time (assuming 1 hour for exercises level 3). After 1 hour I went first to Wikipedia to get some hint before actually reading MacKay's answer. From there I found the integral geometry formulation and the rest was easy. Life in high-dimensional spaces was also quite interesting although I had encountered this idea before when learning about the curse of dimensionality . One thing that helps me think about it is the difference between having uniformly distributed points in a sphere and having points spread in the sphere so that the deviation from the mean distance between points is very small. Putting it more clearly, think about how could you spread points in a circle (I tried triangle first) so that the distance between a point and its closest neighbour is almost the same for every point? Solving and implementing this gave me a bet...

I just finished my first week and want to reflect a bit on why I started this. There are two main reasons: Deepen and broaden my knowledge of Information Theory; and motivate people who might hesitate on taking up the challenge or feel too old to start learning or revisiting a field. The book is very exciting; Mackay's writing is clear and flows nicely, introducing a concept, diving into it, and then out to generalise. His approach is to make you feel as if you are constructing and checking the concepts alongside him. His magic is that the presentation feels both casual and principled, like if you were sitting together with a very smart friend who is explaining you things. I ended this week at page 21, so my current rate is 3 pages per day since I will not be working the weekend. At the current rate I would finish the book in about 6 months. Mackay's prediction for the time required for each exercise worked quite well for me although it took me more time tha...

Hi there, I am a PhD candidate in Computer Vision for Robotics. I use Machine Learning in my research and wanted to get a deeper understanding of the techniques I use. I created this blog to track my progress in going through David Mackay's book Information Theory, Inference, and Learning Algorithms (http://www.inference.phy.cam.ac.uk/mackay/itila/). I will go through every single exercise in the book in chapter order. My intention is to encourage people to study topics that they used to know, but forgot or that they are afraid to go into because of the effort and time required. I believe we should get rid of excuses when we want to learn something. If something deep in you tells that the information in a book or course is essential, you should go for it. After all, you will be better equipped to reflect on the necessity of the learning while you actually go through it. I will post weekly , keeping track of the time spent in each exercise and overall time spent in each ...