umar's coding blog

 The file is a csv structured like this I haven't looked at what the format of the pixels is (i downloaded the months a while ago and cona't remember it's description), whether it's greyscale or rgb.  Also i don't know what the ethnicity numbers indicate, (but it doesn't matter, if i get high accuracy with a model, only then will it relevant). I will be using a 'type-oriented' styel of coding to tackle this (look on my previous post of html/xml to see what i mean). This is important for this reason -  in the past, i have attempted to do feature engineering in PROLOG.It is the most 'natural' way to code i felt, because it is such a great query language. It gives your unlimited flexibility query into a graph (and a graph can represent anything). However, i wold end up getting 'lost' because it's modularity is not so good. After predicates (prolog's name for functions) started piling up, good documentation would have been needed to n...

Few days ago I hacked around a little library for generating web pages in haskell.  I want to talk about it a little because it done in a 'type-oriented' way i.e. i wrote the algebraic datatypes first, and then proceeded with the 'code' that does anything. Having watched a few videos on category theory for hackers, this means i defined objects (in the category theory sense) first, and then wrote morphisms to translate between them (functions in the haskell). I started with the XML type. As can be seen, it's a thing, that contains other things (elements). These things can have some attached key-value pairs (attributes), and also a list of other those things (it's a recursive structure), or a piece of text. In other words, it's a tree, with the leaves being text, and the nodes can have attached key-value pairs. Next the HTML type HTML is also a tree like XML, but has a greater variety of nodes. as listed in the definit of Element. There are things like breaks,...

 I downloaded datase form kaggle. It has age gender, ethnicity labels for photos of people's faces. I downloaded it because it's small and straightforward. I'm going to attempt to see how far accuracy i can get using a combination of straight forward linear regression using vowpal wabbit and feature engineering haskell. I have some ideas I want to try out on it. like  introduction of noise at various levels (it might be what are known as diffusionmodels, though i'm not completely sure thatis what they are).