Go engine

Mastering the game of Go with deep neural networks and tree search

David Silver, Aja Huang1, Chris J. Maddison, Arthur Guez, Laurent Sifre1, George van den Driessche, Julian Schrittwieser, Ioannis Antonoglou, Veda Panneershelvam, Marc Lanctot, Sander Dieleman, Dominik Grewe,
John Nham, Nal Kalchbrenner, Ilya Sutskever, Timothy Lillicrap, Madeleine Leach1, Koray Kavukcuoglu,
Thore Graepel1, Demis Hassabis

The game of Go has long been viewed as the most challenging of classic games for artificial intelligence owing to its enormous search space and the difficulty of evaluating board positions and moves. Here we introduce a new approach to computer Go that uses ‘value networks’ to evaluate board positions and ‘policy networks’ to select moves. These deep neural networks are trained by a novel combination of supervised learning from human expert games, and reinforcement learning from games of self-play. Without any lookahead search, the neural networks play Go at the level of stateof-the-art Monte Carlo tree search programs that simulate thousands of random games of self-play. We also introduce a new search algorithm that combines Monte Carlo simulation with value and policy networks. Using this search algorithm,our program AlphaGo achieved a 99.8% winning rate against other Go programs, and defeated the human European Go champion by 5 games to 0. This is the first time that a computer program has defeated a human professional player in the full-sized game of Go, a feat previously thought to be at least a decade away.
Continue reading “Go engine”

Data explorer


In case you do not live in New York City or you did not attend Data Gotham, do not worry because nearly all the videos and talks are posted on the Data Gotham 2013 Youtube page.

Logan Symposium: Google Public Data Explorer from Berkeley Graduate School of Journalism on FORA.tv

4th Annual Logan Symposium on Investigative Reporting

Read more at http://fora.tv/2010/04/18/Logan_Symposium_Google_Public_Data_Explorer#uXLe1TU6lWF4IJp2.99

Uploaded on Jun 2, 2010

Complete video at: http://fora.tv/2010/04/18/Logan_Sympo…

Using Google’s new Public Data Explorer tool, Ola Rosling demonstrates the effectiveness of visualizing datasets. Looking toward the next political election, Rosling hopes voters will use the tool to answer questions like: How was the money spent? Where are the biggest problems?


Ola Rosling of Google Public Data gives a presentation titled, “Google Public Data Explorer” at the Berkeley Graduate School of Journalism. This program was recorded on April 18, 2010.

Ola Rosling co-founded the Gapminder Foundation and led the development of Trendalyzer, a software that converts time series statistics into animated, interactive and comprehensible graphics. The aim of his work is to promote a fact-based world view through increased use and understanding of freely accessible public data.

In March 2007, Google acquired the Trendalyzer software, where Rosling and his team are now scaling up their tools and making them freely available for any individual or organization to use for analyzing and visualizing data.

DSPL Tools

DSPL Tools is a small suite of command-line utilities designed to help generate, organize, and validate DSPL datasets. The suite currently includes the following components:

  • DSPL Check: Checks a dataset against a variety of criteria including adherence to the official DSPL schema, consistency of internal references, and CSV layout.
  • DSPL Gen: Generates a simple, DSPL dataset “template” from an input CSV file

This software is released under a BSD license; the full source code is available for browsing and download on the DSPL open source site. Release notes are provided in the DSPL Tools README file.

DSPL Developer Guide

DSPL stands for Dataset Publishing Language. It is a representation format for both the metadata (information about the dataset, such as its name and provider, as well as the concepts it contains and displays) and actual data of datasets. Datasets described in this format can be imported into the Google Public Data Explorer, a tool that allows for rich, visual exploration of the data.

Note: To upload data to Google Public Data using the Public Data upload tool, you must have a Google Account.

This document is intended for data owners who want their content to be available in the Public Data Explorer. It goes beyond the Tutorial by diving deeper into the details of the DSPL schema and supported features. Only a basic familiarity of XML is assumed, although knowledge of relational databases is also useful.

Although not a requirement, we suggest reading through the Tutorial, which is shorter and easier to digest, before looking at this document.

Dalvik VM Internals

Dan Bornstein (Google)

Dalvik — the virtual machine with the unusual name — runs your code on Android. Join us to learn about the motivation for its design and get
some details about how it works. You’ll also walk away with a few tips for how to write code that works well with the platform. Be prepared
for a deep dive into technical details. Questions encouraged!

Presentation Slides

Search Engine Optimization (SEO)

Internet es la calle más transitada del mundo, pero el trafico en cada pagina depende principalmente del posicionamiento en los buscadores como Google. Al arte de colocarse en los primeros lugares de los listados se le conoce como Search Engine Optimization (SEO).

Google mantiene como secreto la mecánica de asignación de lugares, que además cambia de manera continúa. Es un proceso bastante errático, y los que logran colocarse en la primera pagina para la lista de búsqueda de un conjunto de palabras, tenderán a mantenerse ahí hagan lo que hagan, tengan el contenido que tengan, siempre y cuando Google no los vete, por razones también erráticas y misteriosas. Es decir, el SEO es un deporte extremo.

Referencias, recursos, y ejemplos