Week 5 Readings

Wikipedia: Data Compression

I really enjoyed this read, partly because I wasn't utterly lost reading it!

When I read about how "lossless compression algorithms usually exploit statistical redundancy", and later: "any compression algorithm will necessarily fail to compress any data containing no discernible patterns", I first generalized that these types of algorithms could be used to compress language (words) but not number data. But in actuality, numerical data is by no means random. It should even be predictable to some extent, depending on what it is describing. The article mentioned the ability to find patterns in numbers themselves, such as repeating digits. But are there algorithms that recognize the probability of certain patterns occuring for data that describes weather vs. test scores vs. race times, for example?

It also caught my attention to find that the study of rate-distortion theory was so old - dating to the very earliest days of the computer era - and the linked articles about Claude Shannon were especially interesting.

Data Compression Basics

I had so many "aha" moments reading this article! I am pretty remedial when it comes to computer matters, but readings like this help establish so many new connections in my brain. For instance, I knew nothing about what file extensions meant, except for the frustration they could cause me when trying to open them with incompatible programs. Now I understand that the extension basically indicates how the file is to be decompressed, which explains the ghastly stings of unreadable code that can appear when I open a file with the wrong application (if the file opens at all)

This also helped to really break down what audio, stills, and video data looks like at a fundamental level, and therefore why so many different compression methods are needed.

By far the most informative and enjoyable read so far - I even liked the writing style and subtle British humor.


Other articles and Muddiest Point to follow...