Trivial or WrongMachine Learning, Data Science, Programming, Mathematics and Algorithms.
http://omersha.github.io//
Thu, 09 Mar 2017 20:27:21 +0000Thu, 09 Mar 2017 20:27:21 +0000Jekyll v3.4.1A Whole New Dimension of Optimization<h1 id="a-whole-new-dimension-of-optimization">A Whole New Dimension of Optimization</h1>
Tue, 10 Jan 2017 00:00:00 +0000
http://omersha.github.io//2017/01/10/a-whole-new-dimension-of-optimization.html
http://omersha.github.io//2017/01/10/a-whole-new-dimension-of-optimization.htmlTake the Rough With the Smooth<div><p>
The starting point of the <a href="http://www.trivialorwrong.com/2016/11/15/the-
roots-of-no-evil.html">previous post</a> was optimization of differentiable
functions, but trying to utilize Fermat's theorem led us eventually to
bracketing algorithms, which make no use of the derivative and are applicable for
non-differentiable functions.
</p></div>
Mon, 28 Nov 2016 00:00:00 +0000
http://omersha.github.io//2016/11/28/take-the-rough-with-the-smooth.html
http://omersha.github.io//2016/11/28/take-the-rough-with-the-smooth.htmlThe Roots of No Evil<blockquote>
<p>“…nothing at all takes place in the universe in which some rule of maximum
or minimum does not appear…” -Leonhard Euler</p>
</blockquote>
Tue, 15 Nov 2016 00:00:00 +0000
http://omersha.github.io//2016/11/15/the-roots-of-no-evil.html
http://omersha.github.io//2016/11/15/the-roots-of-no-evil.htmlRandom Integers<div><p>
The need to sample from a discrete distribution given by a probability vector
$\vec{p}=(p_1,p_2,...,p_n)$ comes up very often. For example, when taking a
bootstrapped sample from a weighted dataset, or when selecting parents from a
population based on their realtive fitness in genetic algorithms, or when
implementing an agent that applies a mixed strategy, or... well, I made my
point: often.
</p></div>
Mon, 26 Sep 2016 00:00:00 +0000
http://omersha.github.io//2016/09/26/random-integers.html
http://omersha.github.io//2016/09/26/random-integers.htmlMachine Learning. Literally.<p>What’s the most successful application of machine learning?</p>
Wed, 08 Jun 2016 00:00:00 +0000
http://omersha.github.io//2016/06/08/machine-learning-literally.html
http://omersha.github.io//2016/06/08/machine-learning-literally.htmlLaws, Sausages and ConvNets<blockquote>
<p>Laws, like sausages, cease to inspire respect in proportion as we know how
they are made.</p>
</blockquote>
Wed, 01 Jun 2016 00:00:00 +0000
http://omersha.github.io//2016/06/01/laws-sausages-and-convnets.html
http://omersha.github.io//2016/06/01/laws-sausages-and-convnets.htmlThe Generative-Discriminative Fallacy<h3 id="1-the-generative-discriminative-dichotomy">1. The Generative-Discriminative Dichotomy</h3>
<div><p>
It is customary to categorize predictive algorithms as either <a href="https://en.wikipedia.org/wiki/Generative_model">"generative"</a>
or <a href="https://en.wikipedia.org/wiki/Discriminative_model">"discriminative"</a>.
Denoting by $Y$ the target and by $X$ the observable, discriminative algorithms are those
that model the conditional distribution $P(Y|X;\theta)$ and generative
algorithms are those that model the joint distribution $P(X,Y|\theta)$) (note
that this entire discussion presumes a probabilistic perspective on predictive
tasks).
</p></div>
Sun, 22 May 2016 00:00:00 +0000
http://omersha.github.io//2016/05/22/the-generative-discriminative-fallacy.html
http://omersha.github.io//2016/05/22/the-generative-discriminative-fallacy.htmlLearning Dynamical Systems<h3 id="1-motivational-speech">1. Motivational Speech</h3>
Sat, 14 May 2016 00:00:00 +0000
http://omersha.github.io//2016/05/14/learning-dynamical-systems.html
http://omersha.github.io//2016/05/14/learning-dynamical-systems.htmlThe Name of The Rose<div><p>
Many ideas in machine learning came up independently in different contexts, and
it's not uncommon to have multiple terms for the same concept. Those instances
are usually dismissed as nuisance; after all, a rose by any other name, et
cetera. Indeed, often this is just a matter of nomenclature, and with time
conventions form, and some terms disappear while others acquire universal
meaning. But sometimes the distinction is not as superficial as it may seem.
</p></div>
Tue, 22 Dec 2015 00:00:00 +0000
http://omersha.github.io//2015/12/22/the-name-of-the-rose.html
http://omersha.github.io//2015/12/22/the-name-of-the-rose.htmlPointless Topology via Abstract Nonsense<div><p>
In 1940. G.H. Hardy published his essay "A Mathematician's Apology" in which he
used number theory as a prominent example for pure and inapplicable mathematics.
He meant it seriously, not as a joke. To the defense of his apology (pun a bit
intended), at the time only a handful of programmers thought about hash-tables,
cryptosystems were yet to be widely implemented in E-commerce platforms, and not
many people used FFT before 1965 (no one but Gauss, I think). Indeed, computers
forced mathematicians to work harder and harder in order to keep annoying
engineers away from their offices.
</p></div>
Tue, 15 Dec 2015 00:00:00 +0000
http://omersha.github.io//2015/12/15/pointless-topology-via-abstract-nonsense.html
http://omersha.github.io//2015/12/15/pointless-topology-via-abstract-nonsense.html