recentpopularlog in


« earlier   
A general reinforcement learning algorithm that masters chess, shogi, and Go through self-play | Science
The game of chess is the longest-studied domain in the history of artificial intelligence. The strongest programs are based on a combination of sophisticated search techniques, domain-specific adaptations, and handcrafted evaluation functions that have been refined by human experts over several decades. By contrast, the AlphaGo Zero program recently achieved superhuman performance in the game of Go by reinforcement learning from self-play. In this paper, we generalize this approach into a single AlphaZero algorithm that can achieve superhuman performance in many challenging games. Starting from random play and given no domain knowledge except the game rules, AlphaZero convincingly defeated a world champion program in the games of chess and shogi (Japanese chess), as well as Go.
videogames  AI  research  algorithm 
2 days ago by basemaly
axiomhq/quantiles: Optimal Quantile Approximation in Streams
This is a translation of TensorFlow's quantile helper class, it aims to compute approximate quantiles with error bound guarantees for weighted data sets. This implementation is an adaptation of techniques from the following papers:
algorithm  library  analytics 
3 days ago by janpeuker
Show HN: Stock Trading from Google Spreadsheet | Hacker News
"I worked for one of the largest financials services companies in the world and they STILL use excel to drive their trading activities. They are so large and so complex that it's impossible to convert them because no one completely understands how it all works. They were developed by traders/market experts with no help from IT, for years IT didn't even knew they existed."

What could *possibly* go wrong?
notes  linux  algorithm 
4 days ago by keithpeter
Algorithms for Historical Maps
In our research, we invented a tool that assigns place markers to labels on old maps. What is that good for? Well, as you can see above, historical maps often contain a surprising amount of labeled places. This is interesting content one can extract from the map. via Pocket
algorithm  analysis  history  maps  metadata 
4 days ago by kintopp

Copy this bookmark:

to read