 |
|
|
|
|
|
|
|
|
|
|
|
|
 |
|
|||||||||||
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
|
|
 |
 |
|
||||||||||
 |
|
|||||||||||
 |
|
|||||||||||
 |
 |
|
||||||||||
| Seminar Schedule (Seminar Organizer: Prof. Ziv Bar-Joseph) | |
Machine Learning Lunchtime Chats
|
Spring 2008 Seminar
Date: May 12, 2008 Time: 4:30 PM - 5:30 PM (Refreshments at 4:15) Location: 1305 Newell-Simon Hall Speaker: Jason Eisner Associate Professor, Johns Hopkins University Title: Weighted Deduction as a Programming Language Abstract: The field of AI has become implementation-bound. Many of us find that considerable research time is lost to software engineering. It is slow work to architect systems that can correctly test our ideas (models and algorithms) on large-scale, incomplete or noisy data. This situation presents barriers to entry, to education, and to modifying and combining existing ideas. In this talk, I'll propose a new level of abstraction, in the form of a programming language. Dyna is a declarative language that combines logic programming with functional programming. It may be regarded as a kind of deductive database, theorem prover, truth maintenance system, or equation solver. I will illustrate how Dyna makes it easy to specify the combinatorial structure of typical computations needed in natural language processing, machine learning, and elsewhere in AI. Then I will sketch implementation strategies and program transformations that can help to make these computations fast and memory-efficient. Finally, I will suggest that machine learning should be used to search for the right strategies for a program on a particular workload. Speaker Bio: Jason Eisner is Associate Professor of Computer Science at Johns Hopkins University, where he is also affiliated with the Center for Language and Speech Processing, the Cognitive Science Department, and the national Center of Excellence in Human Language Technology. He recently served as program chair of the large EMNLP conference. He is particularly interested in designing algorithms that statistically exploit linguistic structure. His 60 or so papers have presented a number of parsing and MT algorithms; finite-state algorithms; formalizations, algorithms, theorems and empirical results in computational phonology; and unsupervised learning methods for various problems including shallow and deep syntax induction.
