### Abstract

We consider a variant of the 'population learning model' proposed by Kearns and Seung, in which the learner is required to be 'distribution-free' as well as computationally efficient. A population learner receives as input hypotheses from a large population of agents and produces as output its final hypothesis. Each agent is assumed to independently obtain labeled sample for the target concept and outputs a hypothesis. A polynomial time population learner is said to 'PAC learn' a concept class, if its hypothesis is probably approximately correct whenever the population size exceeds a certain bound which is polynomial, even if the sample size for each agent is fixed at some constant. We exhibit some general population learning strategies, and some simple concept classes that can be learned by them. These strategies include the 'supremum hypothesis finder,' the 'minimum superset finder' (a special case of the 'supremum hypothesis finder'), and various voting schemes. When coupled with appropriate agent algorithms, these strategies can learn a variety of simple concept classes, such as the 'high-low game,' conjunctions, axis-parallel rectangles and others. We give upper bounds on the required population size for each of these cases, and show that these systems can be used to obtain a speed up from the ordinary PAC-learning model, with appropriate choices of sample and population sizes. With the population learner restricted to be a voting scheme, what we have is effectively a model of 'population prediction,' in which the learner is to predict the value of the target concept at an arbitrarily drawn point, as a threshold function of the predictions made by its agents on the same point. We show that the population learning model is strictly more powerful than the population prediction model. Finally we consider a variant of this model with classification noise, and exhibit a population learner for the class of conjunctions in this model.

Original language | English |
---|---|

Title of host publication | Algorithmic Learning Theory - 4th International Workshop on Analogical and Inductive Inference, AII 1994 and 5th International Workshop on Algorithmic Learning Theory, ALT 1994, Proceedings |

Publisher | Springer Verlag |

Pages | 500-515 |

Number of pages | 16 |

Volume | 872 LNAI |

ISBN (Print) | 9783540585206 |

Publication status | Published - Jan 1 1994 |

Externally published | Yes |

Event | 4th International Workshop on Analogical and Inductive Inference, AII 1994 and 5th International Workshop on Algorithmic Learning Theory, ALT 1994 - Reinhardsbrunn Castle, Germany Duration: Oct 10 1994 → Oct 15 1994 |

### Publication series

Name | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
---|---|

Volume | 872 LNAI |

ISSN (Print) | 0302-9743 |

ISSN (Electronic) | 1611-3349 |

### Other

Other | 4th International Workshop on Analogical and Inductive Inference, AII 1994 and 5th International Workshop on Algorithmic Learning Theory, ALT 1994 |
---|---|

Country | Germany |

City | Reinhardsbrunn Castle |

Period | 10/10/94 → 10/15/94 |

### Fingerprint

### All Science Journal Classification (ASJC) codes

- Theoretical Computer Science
- Computer Science(all)

### Cite this

*Algorithmic Learning Theory - 4th International Workshop on Analogical and Inductive Inference, AII 1994 and 5th International Workshop on Algorithmic Learning Theory, ALT 1994, Proceedings*(Vol. 872 LNAI, pp. 500-515). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 872 LNAI). Springer Verlag.

**Efficient distribution-free population learning of simple concepts.** / Nakamura, Atsuyoshi; Abe, Naoki; Takeuchi, Junnichi.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

*Algorithmic Learning Theory - 4th International Workshop on Analogical and Inductive Inference, AII 1994 and 5th International Workshop on Algorithmic Learning Theory, ALT 1994, Proceedings.*vol. 872 LNAI, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 872 LNAI, Springer Verlag, pp. 500-515, 4th International Workshop on Analogical and Inductive Inference, AII 1994 and 5th International Workshop on Algorithmic Learning Theory, ALT 1994, Reinhardsbrunn Castle, Germany, 10/10/94.

}

TY - GEN

T1 - Efficient distribution-free population learning of simple concepts

AU - Nakamura, Atsuyoshi

AU - Abe, Naoki

AU - Takeuchi, Junnichi

PY - 1994/1/1

Y1 - 1994/1/1

N2 - We consider a variant of the 'population learning model' proposed by Kearns and Seung, in which the learner is required to be 'distribution-free' as well as computationally efficient. A population learner receives as input hypotheses from a large population of agents and produces as output its final hypothesis. Each agent is assumed to independently obtain labeled sample for the target concept and outputs a hypothesis. A polynomial time population learner is said to 'PAC learn' a concept class, if its hypothesis is probably approximately correct whenever the population size exceeds a certain bound which is polynomial, even if the sample size for each agent is fixed at some constant. We exhibit some general population learning strategies, and some simple concept classes that can be learned by them. These strategies include the 'supremum hypothesis finder,' the 'minimum superset finder' (a special case of the 'supremum hypothesis finder'), and various voting schemes. When coupled with appropriate agent algorithms, these strategies can learn a variety of simple concept classes, such as the 'high-low game,' conjunctions, axis-parallel rectangles and others. We give upper bounds on the required population size for each of these cases, and show that these systems can be used to obtain a speed up from the ordinary PAC-learning model, with appropriate choices of sample and population sizes. With the population learner restricted to be a voting scheme, what we have is effectively a model of 'population prediction,' in which the learner is to predict the value of the target concept at an arbitrarily drawn point, as a threshold function of the predictions made by its agents on the same point. We show that the population learning model is strictly more powerful than the population prediction model. Finally we consider a variant of this model with classification noise, and exhibit a population learner for the class of conjunctions in this model.

AB - We consider a variant of the 'population learning model' proposed by Kearns and Seung, in which the learner is required to be 'distribution-free' as well as computationally efficient. A population learner receives as input hypotheses from a large population of agents and produces as output its final hypothesis. Each agent is assumed to independently obtain labeled sample for the target concept and outputs a hypothesis. A polynomial time population learner is said to 'PAC learn' a concept class, if its hypothesis is probably approximately correct whenever the population size exceeds a certain bound which is polynomial, even if the sample size for each agent is fixed at some constant. We exhibit some general population learning strategies, and some simple concept classes that can be learned by them. These strategies include the 'supremum hypothesis finder,' the 'minimum superset finder' (a special case of the 'supremum hypothesis finder'), and various voting schemes. When coupled with appropriate agent algorithms, these strategies can learn a variety of simple concept classes, such as the 'high-low game,' conjunctions, axis-parallel rectangles and others. We give upper bounds on the required population size for each of these cases, and show that these systems can be used to obtain a speed up from the ordinary PAC-learning model, with appropriate choices of sample and population sizes. With the population learner restricted to be a voting scheme, what we have is effectively a model of 'population prediction,' in which the learner is to predict the value of the target concept at an arbitrarily drawn point, as a threshold function of the predictions made by its agents on the same point. We show that the population learning model is strictly more powerful than the population prediction model. Finally we consider a variant of this model with classification noise, and exhibit a population learner for the class of conjunctions in this model.

UR - http://www.scopus.com/inward/record.url?scp=0346926398&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0346926398&partnerID=8YFLogxK

M3 - Conference contribution

SN - 9783540585206

VL - 872 LNAI

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 500

EP - 515

BT - Algorithmic Learning Theory - 4th International Workshop on Analogical and Inductive Inference, AII 1994 and 5th International Workshop on Algorithmic Learning Theory, ALT 1994, Proceedings

PB - Springer Verlag

ER -