3月18日：LLCC讲座An Introduction to a new learning method
来源：会议讲座
作者：
时间：2016-03-16
3月18日周五LLCC讲座
标题： An Introduction to a new learning method
主讲人：孙溢凡 （物理学院）
主持人：周北海教授（哲学系）
评论人：吴玺宏教授 （智能科学系）
时间：2016-03-18 周五 15:10:10
地点：北京大学人文学苑2号楼（哲学系）地下B114
简介：
除了深度学习之外，近期机器学习领域的一个热点是probabilistic program induction。本次语言、逻辑、认知、计算论坛（LLCC）将围绕这一新工作展开讨论。如果大家对AlphaGo背后的深度学习以及蒙特卡洛树搜索感兴趣也欢迎参加，可以与吴玺宏老师讨论。
Abstract: People learning new concepts can often generalize successfully from just a single example, yet machine learning algorithms typically require tens or hundreds of examples to perform with similar accuracy.
In December last year a paper titled “Human-level concept learning through probabilistic program induction” by Lake et. al. was published as a Science cover story. In this paper, the author presented a computational model that captures some concept learning abilities of human in handwritten characters. The so-called hierarchical bayesian program learning (HBPL) model represents concepts as simple programs that best explain observed examples under a Bayesian criterion. On a challenging one-shot classification task, the model achieves human-level performance while outperforming recent deep learning approaches.
原文：Lake, B. M., Salakhutdinov, R., and Tenenbaum, J. B. (2015). Human-level concept learning through probabilistic program induction. Science, 350(6266), 1332-1338.
北京大学 语言、逻辑、认知与计算论坛
llcc.pku.edu.cn
标题： An Introduction to a new learning method
主讲人：孙溢凡 （物理学院）
主持人：周北海教授（哲学系）
评论人：吴玺宏教授 （智能科学系）
时间：2016-03-18 周五 15:10:10
地点：北京大学人文学苑2号楼（哲学系）地下B114
简介：
除了深度学习之外，近期机器学习领域的一个热点是probabilistic program induction。本次语言、逻辑、认知、计算论坛（LLCC）将围绕这一新工作展开讨论。如果大家对AlphaGo背后的深度学习以及蒙特卡洛树搜索感兴趣也欢迎参加，可以与吴玺宏老师讨论。
Abstract: People learning new concepts can often generalize successfully from just a single example, yet machine learning algorithms typically require tens or hundreds of examples to perform with similar accuracy.
In December last year a paper titled “Human-level concept learning through probabilistic program induction” by Lake et. al. was published as a Science cover story. In this paper, the author presented a computational model that captures some concept learning abilities of human in handwritten characters. The so-called hierarchical bayesian program learning (HBPL) model represents concepts as simple programs that best explain observed examples under a Bayesian criterion. On a challenging one-shot classification task, the model achieves human-level performance while outperforming recent deep learning approaches.
原文：Lake, B. M., Salakhutdinov, R., and Tenenbaum, J. B. (2015). Human-level concept learning through probabilistic program induction. Science, 350(6266), 1332-1338.
北京大学 语言、逻辑、认知与计算论坛
llcc.pku.edu.cn