Penny Arcade, Extra Credits, and Stack Exchange Explain What a Developer Does
Penny Arcade, Extra Credits, and Stack Exchange came together to explain what a developer really does.
Want to know what Python web hosting should be like? Take a look at Heroku
I haven’t even fully exposed myself to all of Heroku yet but I am really liking what I am seeing.
Total Probability
Total probability is a very useful way to inspect the probability of a certain state when you do not have the probability value available. Whenever a formula requires the probability of a state, total probability provides a possible way to come up with that value. This situation can come up for instance in the denominator [...]
A Simplified Explanation of Machine Learning
A pretty good video explaining the mechanics of machine learning.
Kernel Functions for Support Vector Machines
Here is a nice collection of kernel functions for use in Support Vector Machines.
Hidden Markov Models (HMM)
Hidden markov models are used to analyze and predict states in a time series.
This is a very useful sub-field in AI because HMM and filters are used in many places including robotics, medical, finance, and speech recognition. It’s a very good bet that Google’s voice commands and Apple’s Siri use HMM and filters.
Blog Listings Now With 90% Less Content!
The blog theme has been adjusted to show excerpts instead of the entire post when being listed with other entries. It will be much easier to browse the site now.
Reinforcement Learning: Active Reinforcement Learning
Active Reinforcement Learning involves re-evaluating a policy $\pi$ to see if there can be better policies. A greedy agent looks at the utility derived from the temporal difference learning process and possibly chooses a better policy. In the TD link I used a deterministic example but it is very possible that a stochastic environment could have caused [...]
Reinforcement Learning: Q Learning
Q Learning is similar to Temporal Difference Learning with a slight distinction and modification to the algorithm. The utility of the state is not learned directly. We look at the Q-Utility, which is the value of the state combined with the action to get there We don’t need the transition model (stochastic probability). This is [...]
Reinforcement Learning: Passive Temporal Difference Learning
Temporal Difference Learning (TD Learning) is a straightforward passive reinforcement learning algorithm. Some key rules are: It always follows a fixed policy $\pi$ that never changes Build a table of utilities $U[s]$ for each state. A utility is a score that measures the value for being in that state, and the learning algorithm is used [...]