22
Research Research Prediction Games in Infinitely Rich Worlds Omid Madani Yahoo! Research

Research Prediction Games in Infinitely Rich Worlds Omid Madani Yahoo! Research

Embed Size (px)

Citation preview

Page 1: Research Prediction Games in Infinitely Rich Worlds Omid Madani Yahoo! Research

ResearchResearch

Prediction Games in Infinitely Rich Worlds

Omid Madani

Yahoo! Research

Page 2: Research Prediction Games in Infinitely Rich Worlds Omid Madani Yahoo! Research

ResearchResearch

“Rather, the formation and use of categories is the stuff of experience.”

Philosophy in the Flesh, Lakoff and Johnson.

Page 3: Research Prediction Games in Infinitely Rich Worlds Omid Madani Yahoo! Research

ResearchResearch

Motivation

• Higher intelligence requires myriad inter-related categories

• How can such be acquired?• Programming them unlikely to be

successful:• Limits of our explicit knowledge• Unknown/unfamiliar domains• Making the system operational..

Page 4: Research Prediction Games in Infinitely Rich Worlds Omid Madani Yahoo! Research

ResearchResearch

Learn? … How?

• “Supervised” learning likely inadequate:• Required:

• ~millions of categories and beyond..• Billions of weights, and beyond..

• Inaccessible “knowledge” (see last slide!)

• Other approaches are fall short (incomplete, etc): clustering, RL, active learning, etc..

Page 5: Research Prediction Games in Infinitely Rich Worlds Omid Madani Yahoo! Research

ResearchResearch

This Work: An Exploration

• An avenue: “prediction games in infinitely rich worlds”

• Exciting part: • World provides unbounded learning

opportunity! (world is the teacher!)• World enjoys many regularities (e.g.

“hierarchical”)

Page 6: Research Prediction Games in Infinitely Rich Worlds Omid Madani Yahoo! Research

ResearchResearch

This Work• Describe the setting

• The games, categories, …

• Discuss:• Desiderata/constraints• Some of the many

challenges/problems

• Preliminary system/observations..

Page 7: Research Prediction Games in Infinitely Rich Worlds Omid Madani Yahoo! Research

ResearchResearch

The Game

• Repeat • Hide part(s) of the stream• Predict (use context)• Update• Move on

• Goal: predict better ... (subject to constraints)• In the process: categories at different levels of

abstraction learned• Some details: what parts to hide? How much

context? What order?

Page 8: Research Prediction Games in Infinitely Rich Worlds Omid Madani Yahoo! Research

ResearchResearch

In a Nutshell

Prediction System

…. 0011101110000….

After a While

predict observe & update

Prediction System

observe & updatepredict

low level categories

higher level categories(bigger chunks)(bits, characters, edges,…)

(e.g. words, digits, phrases, phone numbers, faces, visual objects, home pages, sites,…)

Page 9: Research Prediction Games in Infinitely Rich Worlds Omid Madani Yahoo! Research

ResearchResearch

Example of Games (text)

• .. d?an.. • System predictions (ranked or

assigned probabilities, or.. )• “r”• “e”• “o”• …

• I ? my bike to school.

Page 10: Research Prediction Games in Infinitely Rich Worlds Omid Madani Yahoo! Research

ResearchResearch

Categories

• Building blocks of intelligence?• Patterns that frequently occur

• External • Internal..• Useful for predicting other categories!• They can have structure/regularities

1. Composition (~conjunctions) of others

2. Grouping (~disjunctions)

Page 11: Research Prediction Games in Infinitely Rich Worlds Omid Madani Yahoo! Research

ResearchResearch

Categories

• Low level examples: 0 and 1 or characters• Provided to the system

• Higher levels:• Sequence of k bits• Words• Phrases• Regular expressions • Phone number, contact info, resume, ...

Page 12: Research Prediction Games in Infinitely Rich Worlds Omid Madani Yahoo! Research

ResearchResearch

Prediction Objective

• Desirable: learn higher level categories (bigger/abstract categories are useful externally)

• Question: how does this relate to improving predictions?

1. Higher level categories improve “context” and can save memory

2. Bigger, save time in playing the game (categories are atomic)

Page 13: Research Prediction Games in Infinitely Rich Worlds Omid Madani Yahoo! Research

ResearchResearch

Goal (evaluation criterion)

• Number of bits (characters) correctly predicted per

unit time (or per prediction action)

• Subject to constraints (space, time,..)

• How about entropy/perplexity? Categories are structured..

Page 14: Research Prediction Games in Infinitely Rich Worlds Omid Madani Yahoo! Research

ResearchResearch

Desiderata/Challenges/Issues• Lots of data!

• Efficiency: space and time!

• Noise:• Statistical insignificance• Significance, but for short time..

• Variety (need for abstraction)• Drift (e.g. developments within system)• Motivate: (primarily) online

algorithms/systems

Page 15: Research Prediction Games in Infinitely Rich Worlds Omid Madani Yahoo! Research

ResearchResearch

Desiderata/Challenges• Why need for “system”s?

• Multiple algorithms/parts needed• Persistence

• Long term learning: how can we make sure noise/errors do not accumulate?

• Control of the input stream..

Page 16: Research Prediction Games in Infinitely Rich Worlds Omid Madani Yahoo! Research

ResearchResearch

Why Now?

• Many category learning is possible/efficient!• Online• Noise tolerant

• Expectation: other problems are solvable..

Page 17: Research Prediction Games in Infinitely Rich Worlds Omid Madani Yahoo! Research

ResearchResearch

Preliminary Report

• Work in Progress!

• Plays the game in text

• Begins at character level

• No segmentation, just a stream

• Makes and predicts larger sequences (composition)

Page 18: Research Prediction Games in Infinitely Rich Worlds Omid Madani Yahoo! Research

ResearchResearch

Preliminary Observations

• Ran on Reuters RCV1 (text body) ( simply zcat dir/file* )

• 800k articles• >= 150 million learning/prediction episodes• Over 10 million categories built• 3-4 hours each pass

Page 19: Research Prediction Games in Infinitely Rich Worlds Omid Madani Yahoo! Research

ResearchResearch

Observations• Performance on held out (one of the

Reuters files):• 8-9 characters long to predict on average• Almost two characters correct on

average, per prediction action

• Can overfit/memorize! (long categories)

• Current: stop category generation in first pass

Page 20: Research Prediction Games in Infinitely Rich Worlds Omid Madani Yahoo! Research

ResearchResearch

Page 21: Research Prediction Games in Infinitely Rich Worlds Omid Madani Yahoo! Research

ResearchResearch

Current/Future

• Much work:• Learn groupings• Recognize/use “syntactic”

categories?• Prediction objective is ok?• Category generation.. What’s a good

method?

• Compare: language modeling, etc

Page 22: Research Prediction Games in Infinitely Rich Worlds Omid Madani Yahoo! Research

ResearchResearch

Much Related Work!

• Online learning, clustering, deep learning, Bayesian methods, hierarchical learning, importance of predictions (“On Intelligence”, “natural computations”), models of neocortex (“circuits of the mind”), concepts (“big book of concepts”), cumulative learning, neural nets, compression, learning an index of categories!