Dynamic Programming for Linear Time Incremental Parsing
|
|
- Della Richard
- 5 years ago
- Views:
Transcription
1 Dynamic Programming for Linear Time ncremental Parsing Liang Huang nformation Sciences nstitute University of Southern California Kenji Sagae nstitute for Creative Technologies University of Southern California ACL 2010, Uppsala, Sweden, July 2010 (slightly expanded)
2 DP for ncremental Parsing (Huang and Sagae) Ambiguities in Parsing feed cats nearby in the garden... let s focus on dependency structures for simplicity ambiguous attachments of nearby and in ambiguity explodes exponentially with sentence length must design efficient (polynomial) search algorithm typically using dynamic programming (DP); e.g. CKY 2
3 DP for ncremental Parsing (Huang and Sagae) Ambiguities in Parsing feed cats nearby in the garden... let s focus on dependency structures for simplicity ambiguous attachments of nearby and in ambiguity explodes exponentially with sentence length must design efficient (polynomial) search algorithm typically using dynamic programming (DP); e.g. CKY 2
4 DP for ncremental Parsing (Huang and Sagae) Ambiguities in Parsing feed cats nearby in the garden... let s focus on dependency structures for simplicity ambiguous attachments of nearby and in ambiguity explodes exponentially with sentence length must design efficient (polynomial) search algorithm typically using dynamic programming (DP); e.g. CKY 2
5 DP for ncremental Parsing (Huang and Sagae) Ambiguities in Parsing feed cats nearby in the garden... let s focus on dependency structures for simplicity ambiguous attachments of nearby and in ambiguity explodes exponentially with sentence length must design efficient (polynomial) search algorithm typically using dynamic programming (DP); e.g. CKY 2
6 DP for ncremental Parsing (Huang and Sagae) 3 But full DP is too slow... feed cats nearby in the garden... full DP (like CKY) is too slow (cubic-time) while human parsing is fast & incremental (linear-time)
7 DP for ncremental Parsing (Huang and Sagae) 3 But full DP is too slow... feed cats nearby in the garden... full DP (like CKY) is too slow (cubic-time) how about incremental parsing then? while human parsing is fast & incremental (linear-time) yes, but only with greedy search (accuracy suffers) explores tiny fraction of trees (even w/ beam search)
8 But full DP is too slow... feed cats nearby in the garden... full DP (like CKY) is too slow (cubic-time) how about incremental parsing then? while human parsing is fast & incremental (linear-time) yes, but only with greedy search (accuracy suffers) explores tiny fraction of trees (even w/ beam search) can we combine the merits of both approaches? a fast, incremental parser with dynamic programming? explores exponentially many trees in linear-time? DP for ncremental Parsing (Huang and Sagae) 3
9 DP for ncremental Parsing (Huang and Sagae) 4 Linear-Time ncremental DP greedy search principled search incremental parsing (e.g. shift-reduce) (Nivre 04; Collins/Roark 04;...) this work: fast shift-reduce parsing with dynamic programming fast (linear-time) full DP (e.g. CKY) (Eisner 96; Collins 99;...) slow (cubic-time)
10 Preview of the Results very fast linear-time dynamic programming parser best reported dependency accuracy on PTB/CTB explores exponentially many trees (and outputs forest) parsing time (secs) sentence length DP for ncremental Parsing (Huang and Sagae) 5
11 Preview of the Results very fast linear-time dynamic programming parser best reported dependency accuracy on PTB/CTB explores exponentially many trees (and outputs forest) parsing time (secs) Charniak Berkeley MST this work sentence length DP for ncremental Parsing (Huang and Sagae) 5
12 Preview of the Results very fast linear-time dynamic programming parser best reported dependency accuracy on PTB/CTB explores exponentially many trees (and outputs forest) parsing time (secs) Charniak number of trees explored Berkeley MST this work sentence length DP for ncremental Parsing (Huang and Sagae) DP: exponential non-dp beam search sentence length 5
13 DP for ncremental Parsing (Huang and Sagae) 6 Outline Motivation ncremental (Shift-Reduce) Parsing Dynamic Programming for ncremental Parsing Experiments
14 Shift-Reduce Parsing feed cats nearby in the garden. action stack queue 0 - feed cats... feed cats nearby... cats nearby in... cats nearby in... nearby in the... nearby in the... in the garden... DP for ncremental Parsing (Huang and Sagae) 7
15 Shift-Reduce Parsing feed cats nearby in the garden. action stack queue 0-1 shift feed cats... feed cats nearby... cats nearby in... cats nearby in... nearby in the... nearby in the... in the garden... DP for ncremental Parsing (Huang and Sagae) 8
16 Shift-Reduce Parsing feed cats nearby in the garden. action stack queue 0-1 shift 2 shift feed feed cats... feed cats nearby... cats nearby in... cats nearby in... nearby in the... nearby in the... in the garden... DP for ncremental Parsing (Huang and Sagae) 9
17 Shift-Reduce Parsing feed cats nearby in the garden. action stack queue 0-1 shift 2 shift 3 l-reduce feed feed feed cats... feed cats nearby... cats nearby in... cats nearby in... nearby in the... nearby in the... in the garden... DP for ncremental Parsing (Huang and Sagae) 10
18 Shift-Reduce Parsing feed cats nearby in the garden. action stack queue 0-1 shift 2 shift 3 l-reduce 4 shift feed feed feed cats feed cats... feed cats nearby... cats nearby in... cats nearby in... nearby in the... nearby in the... in the garden... DP for ncremental Parsing (Huang and Sagae) 11
19 Shift-Reduce Parsing feed cats nearby in the garden. action stack queue 0-1 shift 2 shift 3 l-reduce 4 shift 5a r-reduce feed feed feed cats feed cats feed cats... feed cats nearby... cats nearby in... cats nearby in... nearby in the... nearby in the... in the garden... DP for ncremental Parsing (Huang and Sagae) 12
20 Shift-Reduce Parsing feed cats nearby in the garden. action stack queue 0-1 shift 2 shift 3 l-reduce 4 shift 5a r-reduce 5b shift feed feed feed cats feed cats feed cats nearby feed cats... feed cats nearby... cats nearby in... cats nearby in... nearby in the... nearby in the... in the garden... DP for ncremental Parsing (Huang and Sagae) 13
21 DP for ncremental Parsing (Huang and Sagae) 14 Shift-Reduce Parsing feed cats nearby in the garden. action stack queue 0-1 shift 2 shift 3 l-reduce 4 shift 5a r-reduce 5b shift feed feed feed cats feed cats feed cats nearby shift-reduce conflict feed cats... feed cats nearby... cats nearby in... cats nearby in... nearby in the... nearby in the... in the garden...
22 DP for ncremental Parsing (Huang and Sagae) 15 Choosing Parser Actions stack queue... feed cats in the garden... nearby stack queue... s2 s1 s0 q0 q1... features: (s0.w, s0.rc, q0,...) = (cats, nearby, in,...) score each action using features f and weights w features are drawn from a local window abstraction (or signature) of a state -- this inspires DP! weights trained by structured perceptron (Collins 02)
23 DP for ncremental Parsing (Huang and Sagae) 16 Greedy Search each state => three new states (shift, l-reduce, r-reduce) search space should be exponential greedy search: always pick the best next state
24 DP for ncremental Parsing (Huang and Sagae) 17 Greedy Search each state => three new states (shift, l-reduce, r-reduce) search space should be exponential greedy search: always pick the best next state
25 DP for ncremental Parsing (Huang and Sagae) 18 Beam Search each state => three new states (shift, l-reduce, r-reduce) search space should be exponential beam search: always keep top-b states
26 DP for ncremental Parsing (Huang and Sagae) 19 Dynamic Programming each state => three new states (shift, l-reduce, r-reduce) key idea of DP: share common subproblems merge equivalent states => polynomial space
27 Dynamic Programming each state => three new states (shift, l-reduce, r-reduce) key idea of DP: share common subproblems merge equivalent states => polynomial space graph-structured stack (Tomita, 1988) DP for ncremental Parsing (Huang and Sagae) 20
28 Dynamic Programming each state => three new states (shift, l-reduce, r-reduce) key idea of DP: share common subproblems merge equivalent states => polynomial space graph-structured stack (Tomita, 1988) DP for ncremental Parsing (Huang and Sagae) 21
29 Dynamic Programming each state => three new states (shift, l-reduce, r-reduce) key idea of DP: share common subproblems merge equivalent states => polynomial space each DP state corresponds to exponentially many non-dp states graph-structured stack (Tomita, 1988) DP for ncremental Parsing (Huang and Sagae) 21
30 Dynamic Programming each state => three new states (shift, l-reduce, r-reduce) key idea of DP: share common subproblems merge equivalent states => polynomial space each DP state corresponds to exponentially many non-dp states DP: exponential graph-structured stack (Tomita, 1988) DP for ncremental Parsing (Huang and Sagae) non-dp beam search sentence length 22
31 DP for ncremental Parsing (Huang and Sagae) 23 Merging Equivalent States two states are equivalent if they agree on features because same features guarantee same cost stack queue shift-reduce conflict: feed cats nearby in the garden sh... feed re feed sh... cats... s2 s1 s0 q0 q1... feed cats nearby in the garden
32 DP for ncremental Parsing (Huang and Sagae) 23 Merging Equivalent States two states are equivalent if they agree on features because same features guarantee same cost stack queue shift-reduce conflict: feed cats nearby in the garden sh... feed re feed sh... cats... s2 s1 s0 q0 q1... assume features only look at root of s0 feed cats nearby in the garden
33 DP for ncremental Parsing (Huang and Sagae) 23 Merging Equivalent States two states are equivalent if they agree on features because same features guarantee same cost stack queue shift-reduce conflict: feed cats nearby in the garden sh... feed re feed... cats feed cats nearby in the garden sh... s2 s1 s0 q0 q1... assume features only look at root of s0 two states are equivalent if they agree on root of s0
34 DP for ncremental Parsing (Huang and Sagae) 23 Merging Equivalent States two states are equivalent if they agree on features because same features guarantee same cost stack queue shift-reduce conflict: feed cats nearby in the garden sh... feed re feed... cats feed cats nearby in the garden sh... s2 s1 s0 q0 q1... assume features only look at root of s0 two states are equivalent if they agree on root of s0
35 DP for ncremental Parsing (Huang and Sagae) 24 Merging Equivalent States two states are equivalent if they agree on features because same features guarantee same cost shift-reduce conflict: feed cats nearby in the garden feed cats nearby in the garden cats cats sh re... nearby... feed stack queue... s2 s1 s0 q0 q1...
36 DP for ncremental Parsing (Huang and Sagae) 24 Merging Equivalent States two states are equivalent if they agree on features because same features guarantee same cost shift-reduce conflict: feed cats nearby in the garden feed cats nearby in the garden cats cats sh re... nearby... feed stack queue... s2 s1 s0 q0 q1...
37 DP for ncremental Parsing (Huang and Sagae) 25 Merging Equivalent States two states are equivalent if they agree on features because same features guarantee same cost shift-reduce conflict: feed cats nearby in the garden nearby feed cats nearby in the garden cats cats sh re... nearby... feed stack queue... s2 s1 s0 q0 q1... sh re... cats... nearby
38 DP for ncremental Parsing (Huang and Sagae) 26 Merging Equivalent States two states are equivalent if they agree on features because same features guarantee same cost shift-reduce conflict: feed in the garden feed cats nearby... cats nearby... cats sh re... nearby... feed in the garden stack queue... s2 s1 s0 q0 q1... sh re... cats re... feed... nearby re... feed
39 DP for ncremental Parsing (Huang and Sagae) 26 Merging Equivalent States two states are equivalent if they agree on features because same features guarantee same cost shift-reduce conflict: feed in the garden feed cats nearby... cats nearby... cats sh re... nearby... feed in the garden stack queue... s2 s1 s0 q0 q1... sh re... cats re... feed... nearby re... feed
40 DP for ncremental Parsing (Huang and Sagae) 27 Merging Equivalent States two states are equivalent if they agree on features because same features guarantee same cost shift-reduce conflict: feed in the garden feed cats cats nearby nearby cats in the garden stack queue... s2 s1 s0 q0 q1... sh re... nearby... cats re re sh... feed... nearby re... feed
41 DP for ncremental Parsing (Huang and Sagae) 27 Merging Equivalent States two states are equivalent if they agree on features because same features guarantee same cost shift-reduce conflict: feed in the garden feed cats cats nearby nearby cats in the garden stack queue... s2 s1 s0 q0 q1... sh re... nearby... cats re re sh... feed... nearby re (local) ambiguity-packing!... feed
42 DP for ncremental Parsing (Huang and Sagae) 28 Merging Equivalent States two states are equivalent if they agree on features because same features guarantee same cost shift-reduce conflict: feed in the garden feed cats cats nearby nearby cats sh re in the garden stack queue... s2 s1 s0 q0 q1... re... nearby... cats re... feed sh... feed... nearby re sh... in
43 Merging Equivalent States two states are equivalent if they agree on features because same features guarantee same cost shift-reduce conflict: feed in the garden feed cats cats nearby nearby cats sh re in the garden stack queue... s2 s1 s0 q0 q1... re... nearby... cats re... feed sh... feed... nearby re sh... in DP for ncremental Parsing (Huang and Sagae) graph-structured stack 28
44 Theory: Polynomial-Time DP stack queue... s2 s1 s0 q0 q1... this DP is exact and polynomial-time if features are: a) bounded -- for polynomial time features can only look at a local window b) monotonic -- for correctness (optimal substructure) features should draw no more info from trees farther away from stack top than from trees closer to top both are intuitive: a) always true; b) almost always true DP for ncremental Parsing (Huang and Sagae) 29
45 DP for ncremental Parsing (Huang and Sagae) 30 Theory: Monotonic History related: grammar refinement by annotation (Johnson, 1998) annotate vertical context history (e.g., parent) monotonicity: can t annotate grand-parent without annotating the parent (otherwise DP would fail) our features: left-context history instead of vertical-context similarly, can t annotate s2 without annotating s1 but we can always design minimum monotonic superset grand-parent parent s2 s1 s0 stack
46 DP for ncremental Parsing (Huang and Sagae) 31 Related Work Graph-Structured Stack (Tomita 88): Generalized LR GSS is just a chart viewed from left to right (e.g. Earley 70) this line of work started w/ Lang (1974); stuck since 1990 b/c explicit LR table is impossible with modern grammars general idea: compile CFG parse chart to FSAs (e.g. our beam)
47 Related Work Graph-Structured Stack (Tomita 88): Generalized LR GSS is just a chart viewed from left to right (e.g. Earley 70) this line of work started w/ Lang (1974); stuck since 1990 b/c explicit LR table is impossible with modern grammars general idea: compile CFG parse chart to FSAs (e.g. our beam) We revived and advanced this line of work in two aspects theoretical: implicit LR table based on features merge and split on-the-fly; no pre-compilation needed monotonic feature functions guarantee correctness (new) practical: achieved linear-time performance with pruning 31 DP for ncremental Parsing (Huang and Sagae)
48 Experiments
49 DP for ncremental Parsing (Huang and Sagae) 33 Speed Comparison 5 times faster with the same parsing accuracy DP non-dp time (hours)
50 DP for ncremental Parsing (Huang and Sagae) 34 Correlation of Search and Parsing better search quality <=> better parsing accuracy dependency accuracy DP non-dp average model score
51 Search Space: Exponential number of trees explored DP for ncremental Parsing (Huang and Sagae) DP: exponential non-dp: fixed (beam-width) sentence length 35
52 DP for ncremental Parsing (Huang and Sagae) 36 N-Best / Forest Oracles DP forest oracle (98.15) DP k-best in forest non-dp k-best in beam
53 DP for ncremental Parsing (Huang and Sagae) 37 Better Search => Better Learning DP leads to faster and better learning w/ perceptron
54 DP for ncremental Parsing (Huang and Sagae) 38 Learning Details: Early Updates greedy search: update at first error (Collins/Roark 04) beam search: update when gold is pruned (Zhang/Clark 08) DP search: also update when gold is merged (new!) b/c we know gold can t make to the top again
55 Parsing Time vs. Sentence Length parsing speed (scatter plot) compared to other parsers parsing time (secs) sentence length DP for ncremental Parsing (Huang and Sagae) 39
56 Parsing Time vs. Sentence Length parsing speed (scatter plot) compared to other parsers parsing time (secs) Charniak DP for ncremental Parsing (Huang and Sagae) Berkeley this work MST sentence length 39
57 Parsing Time vs. Sentence Length parsing speed (scatter plot) compared to other parsers parsing time (secs) Charniak O(n 2.5 ) O(n 2.4 ) Berkeley this work DP for ncremental Parsing (Huang and Sagae) 39 MST sentence length O(n 2 ) O(n)
58 DP for ncremental Parsing (Huang and Sagae) Final Results much faster than major parsers (even with Python!) first linear-time incremental dynamic programming parser best reported dependency accuracy on Penn Treebank McDonald et al 05 - MST Koo et al 08 baseline* Zhang & Clark 08 single this work Charniak 00 Petrov & Klein time complexity trees searched 0.12 O(n 2 ) exponential - O(n 4 ) exponential 0.11 O(n) constant 0.04 O(n) exponential 0.49 O(n 2.5 ) exponential 0.21 O(n 2.4 ) exponential
59 DP for ncremental Parsing (Huang and Sagae) Final Results much faster than major parsers (even with Python!) first linear-time incremental dynamic programming parser best reported dependency accuracy on Penn Treebank McDonald et al 05 - MST Koo et al 08 baseline* Zhang & Clark 08 single this work Charniak 00 Petrov & Klein time complexity trees searched 0.12 O(n 2 ) exponential - O(n 4 ) exponential 0.11 O(n) constant 0.04 O(n) exponential 0.49 O(n 2.5 ) exponential 0.21 O(n 2.4 ) exponential
60 Final Results much faster than major parsers (even with Python!) first linear-time incremental dynamic programming parser best reported dependency accuracy on Penn Treebank McDonald et al 05 - MST Koo et al 08 baseline* Zhang & Clark 08 single this work Charniak 00 Petrov & Klein time complexity trees searched 0.12 O(n 2 ) exponential - O(n 4 ) exponential 0.11 O(n) constant 0.04 O(n) exponential 0.49 O(n 2.5 ) exponential 0.21 O(n 2.4 ) exponential DP for ncremental Parsing (Huang and Sagae) *at this ACL: Koo & Collins 10: 93.0 with O(n 4 )
61 Final Results on Chinese also the best parsing accuracy on Chinese Penn Chinese Treebank (CTB 5) all numbers below use gold-standard POS tags Duan et al Zhang & Clark 08 (single) this work word 84.4 non-root root DP for ncremental Parsing (Huang and Sagae)
62 Conclusion greedy search incremental parsing (e.g. shift-reduce) principled search fast (linear-time) full dynamic programming (e.g. CKY) slow (cubic-time) DP for ncremental Parsing (Huang and Sagae) 42
63 DP for ncremental Parsing (Huang and Sagae) 42 Conclusion greedy search principled search incremental parsing (e.g. shift-reduce) linear-time shift-reduce parsing w/ dynamic programming fast (linear-time) full dynamic programming (e.g. CKY) slow (cubic-time)
64 DP for ncremental Parsing (Huang and Sagae) Thank You a general theory of DP for shift-reduce parsing fast, accurate DP parser release coming soon: as long as features are bounded and monotonic future work adapt to constituency parsing (straightforward) other grammar formalisms like CCG and TAG integrate POS tagging into the parser 43
Recurrent neural network grammars. Slide credits: Chris Dyer, Adhiguna Kuncoro
Recurrent neural network grammars Slide credits: Chris Dyer, Adhiguna Kuncoro Widespread phenomenon: Polarity items can only appear in certain contexts Example: anybody is a polarity item that tends to
More informationHeuristic search, A* CS171, Winter 2018 Introduction to Artificial Intelligence Prof. Richard Lathrop. Reading: R&N
Heuristic search, A* CS171, Winter 2018 Introduction to Artificial Intelligence Prof. Richard Lathrop Reading: R&N 3.5-3.7 Outline Review limitations of uninformed search methods Informed (or heuristic)
More informationModeling: Having Kittens
PROBLEM SOLVING Mathematics Assessment Project CLASSROOM CHALLENGES A Formative Assessment Lesson Modeling: Having Kittens Mathematics Assessment Resource Service University of Nottingham & UC Berkeley
More informationGrade: 8. Author: Hope Phillips
Title: Fish Aquariums Real-World Connection: Grade: 8 Author: Hope Phillips BIG Idea: Linear Functions Fish aquariums can be found in homes, restaurants, and businesses. From simple goldfish to exotic
More informationThe online processing of semantic and pragmatic content
The online processing of semantic and pragmatic content LINGUIST510 Brian Dillon Psycholinguistics Comprehension: How do we compute the meaning of a sentence in real time? What are the online computations
More informationPopulation Dynamics: Predator/Prey Teacher Version
Population Dynamics: Predator/Prey Teacher Version In this lab students will simulate the population dynamics in the lives of bunnies and wolves. They will discover how both predator and prey interact
More informationESTIMATING NEST SUCCESS: WHEN MAYFIELD WINS DOUGLAS H. JOHNSON AND TERRY L. SHAFFER
ESTIMATING NEST SUCCESS: WHEN MAYFIELD WINS DOUGLAS H. JOHNSON AND TERRY L. SHAFFER U.S. Fish and Wildlife Service, Northern Prairie Wildlife Research Center, Jamestown, North Dakota 58402 USA ABSTRACT.--The
More informationA Column Generation Algorithm to Solve a Synchronized Log-Truck Scheduling Problem
A Column Generation Algorithm to Solve a Synchronized Log-Truck Scheduling Problem Odysseus 2012 Greg Rix 12 Louis-Martin Rousseau 12 Gilles Pesant 13 1 Interuniversity Research Centre on Enterprise Networks,
More informationAlgebra 3 SAILS. Pacing Guide to make an A in the course = equivalent to 21 ACT math sub-score: SAILS Pacing for Traditional Schedule Module 1
Algebra 3 SAILS What is SAILS? SAILS Seamless Alignment Integrated Learning Support. SAILS is a program developed specifically for students whose ACT is 18 or less. Any student with an ACT score 19 or
More information5 State of the Turtles
CHALLENGE 5 State of the Turtles In the previous Challenges, you altered several turtle properties (e.g., heading, color, etc.). These properties, called turtle variables or states, allow the turtles to
More informationAnimal Speeds Grades 7 12
Directions: Answer the following questions using the information provided. Show your work. If additional space is needed, please attach a separate piece of paper and correctly identify the problem it correlates
More informationNathan A. Thompson, Ph.D. Adjunct Faculty, University of Cincinnati Vice President, Assessment Systems Corporation
An Introduction to Computerized Adaptive Testing Nathan A. Thompson, Ph.D. Adjunct Faculty, University of Cincinnati Vice President, Assessment Systems Corporation Welcome! CAT: tests that adapt to each
More informationBuilding Rapid Interventions to reduce antimicrobial resistance and overprescribing of antibiotics (BRIT)
Greater Manchester Connected Health City (GM CHC) Building Rapid Interventions to reduce antimicrobial resistance and overprescribing of antibiotics (BRIT) BRIT Dashboard Manual Users: General Practitioners
More informationDog Years Dilemma. Using as much math language and good reasoning as you can, figure out how many human years old Trina's puppy is?
Trina was playing with her new puppy last night. She began to think about what she had read in a book about dogs. It said that for every year a dog lives it actually is the same as 7 human years. She looked
More informationBuilding Concepts: Mean as Fair Share
Lesson Overview This lesson introduces students to mean as a way to describe the center of a set of data. Often called the average, the mean can also be visualized as leveling out the data in the sense
More informationLab 6: Energizer Turtles
Lab 6: Energizer Turtles Screen capture showing the required components: 4 Sliders (as shown) 2 Buttons (as shown) 4 Monitors (as shown) min-pxcor = -50, max-pxcor = 50, min-pycor = -50, max-pycor = 50
More informationPopulation Dynamics: Predator/Prey Teacher Version
Population Dynamics: Predator/Prey Teacher Version In this lab students will simulate the population dynamics in the lives of bunnies and wolves. They will discover how both predator and prey interact
More informationCatapult Project (Quadratic Functions)
Project Catapult Trajectory (Quadratic Functions) Project Outline Catapult Project (Quadratic Functions) Project Introduction To go along with other subjects at Mater Academy gearing you towards STEAM,
More information[Boston March for Science 2017 photo Hendrik Strobelt]
[Boston March for Science 2017 photo Hendrik Strobelt] [Boston March for Science 2017] [Boston March for Science 2017] [Boston March for Science 2017] Object Detectors Emerge in Deep Scene CNNs Bolei
More informationUniversity of Pennsylvania. From Perception and Reasoning to Grasping
University of Pennsylvania GRASP LAB PR2GRASP: From Perception and Reasoning to Grasping Led by Maxim Likhachev Kostas Daniilides Vijay Kumar Katherine J. Kuchenbecker Jianbo Shi Daniel D. Lee Mark Yim
More informationLogical Forms. Prof. Sameer Singh CS 295: STATISTICAL NLP WINTER February 16, 2017
Logical Forms Prof. Sameer Singh CS 295: STATISTICAL NLP WINTER 2017 February 16, 2017 Based on slides from Noah Smith, Dan Klein, Tom Kwiatkowski, and everyone else they copied from. Outline Logical Semantics
More informationWhat kind of Theory do we need for English Syntax? Are languages finite? Could we list all the sentences of English?
What kind of Theory do we need for English Syntax? Are languages finite? Could we list all the sentences of English? Some sentences go on and on. Some sentences go on and on and on. Some sentences go on
More informationKing Fahd University of Petroleum & Minerals College of Industrial Management
King Fahd University of Petroleum & Minerals College of Industrial Management CIM COOP PROGRAM POLICIES AND DELIVERABLES The CIM Cooperative Program (COOP) period is an essential and critical part of your
More informationMoving toward formalisation COMP62342
Moving toward formalisation COMP62342 Sean Bechhofer sean.bechhofer@manchester.ac.uk Uli Sattler uli.sattler@manchester.ac.uk (thanks to Bijan Parsia for slides) Previously... We started the knowledge
More informationEating Your Own Dog Food
Eating Your Own Dog Food Andreas Abel 1 1 Department of Computer Science and Engineering Chalmers and Gothenburg University, Sweden 25th Agda Implementor s Meeting (AIM XXI) Teknikparken, Chalmers, Gothenburg,
More informationVeterinary Medical Terminology
Curriculum Outline: Course # Required courses prior to admission Credit hours BIO 0 Principles of Biology I with Lab 4 CHM 0 General Chemistry I with Lab 4 ENG 110 or 111 or 1 Freshman Composition or Composition
More informationModeling and Control of Trawl Systems
Modeling and Control of Trawl Systems Karl-Johan Reite, SINTEF Fisheries and Aquaculture Supervisor: Professor A. J. Sørensen * Advisor: Professor H. Ellingsen * * Norwegian University of Science and Technology
More informationHours of manual cash counting reduced to 12 minutes. John G. Shedd Aquarium, USA
Hours of manual cash counting reduced to 12 minutes John G. Shedd Aquarium, USA ABOUT JOHN G. SHEDD AQUARIUM Shedd Aquarium/Brenna Hernandez Glory s machines are a huge time-saver. I don t think we had
More informationOur class had 2 incubators full of eggs. On day 21, our chicks began to hatch. In incubator #1, 1/3 of the eggs hatched. There were 2 chicks.
Our class had 2 incubators full of eggs. On day 21, our chicks began to hatch. In incubator #1, 1/3 of the eggs hatched. There were 2 chicks. How many eggs were in the incubator before hatching? How many
More informationA Novel Approach For Error Detection And Correction Using Prefix-Adders
A Novel Approach For Error Detection And Correction Using Prefix-Adders B. Naga Jyothi* 1, K.S.N.Murthy 2, K.Srinivasarao 3 *1 PG Student Department of ECE, K.L. University Green fields-522502, AP, India
More informationDesign of a High Speed Adder
Design of a High Speed Adder Aritra Mitra 1, Bhavesh Sharma 2, Nilesh Didwania 3 and Amit Bakshi 4 Aritra.mitra000@gmail.com, Abakshi.ece@gmail.com Abstract In this paper we have compared different addition
More informationComponent Specification NFQ Level 5. Sheep Husbandry 5N Component Details. Sheep Husbandry. Level 5. Credit Value 10
Component Specification NFQ Level 5 Sheep Husbandry 5N20385 1. Component Details Title Teideal as Gaeilge Award Type Code Sheep Husbandry Riar Caorach Minor 5N20385 Level 5 Credit Value 10 Purpose Learning
More informationHeuris'c search, A* CS171, Fall 2016 Introduc'on to Ar'ficial Intelligence Prof. Alexander Ihler. Reading: R&N
Heuris'c search, A* CS171, Fall 2016 Introduc'on to Ar'ficial Intelligence Prof. Alexander Ihler Reading: R&N 3.5-3.7 Outline Review limita'ons of uninformed search methods Informed (or heuris/c) search
More informationNumeracy Practice Tests
KEY STAGE 2 LEVEL 6 TEST A Numeracy Practice Tests CALCULATOR NOT ALLOWED 1 Ramya Marc Chelsea First name Last name School Test Instructions You may not use a calculator to answer any questions in this
More informationDesign of High Speed Vedic Multiplier Using Carry Select Adder with Brent Kung Adder
Design of High Speed Vedic Multiplier Using Carry Select Adder with Brent Kung Adder Kathi Anoosha M.Tech(VLSI&ES), AVN Institute of Engineering and Technology. Sasi Kiran, M.Tech Assistant Professor,
More informationWeek 42: Siamese Network: Architecture and Applications in Visual Object Tracking. Yuanwei Wu
Week 42: Siamese Network: Architecture and Applications in Visual Object Tracking Yuanwei Wu 10-21-2016 1 Outline Siamese Architecture Siamese Applications in Computer Vision Paper review Visual Object
More informationBe Doggone Smart at Work
Be Doggone Smart at Work Safety training for dog bite prevention on the job No part of this demo may be copied or used for public presentation or training purposes. This is a free introductory demo containing
More informationJune 2009 (website); September 2009 (Update) consent, informed consent, owner consent, risk, prognosis, communication, documentation, treatment
GUIDELINES Informed Owner Consent Approved by Council: June 10, 2009 Publication Date: June 2009 (website); September 2009 (Update) To Be Reviewed by: June 2014 Key Words: Related Topics: Legislative References:
More informationA Peek Into the World of Streaming
A Peek Into the World of Streaming What s Streaming? Data Stream processing engine Summarized data What s Streaming? Data Stream processing engine Summarized data Data storage Funny thing: Streaming in
More information) the monarch butterfly Reading Behavior Recording Mark Score Accurate Reading Correct / no error Substitution Omission of word Insertion of word Rereads a word, sentence or phrase Child says
More informationEgg laying vs. Live Birth
Egg laying vs. Live Birth Grade Level: This lesson is designed for a 4 th grade class. Science Concept: Animals have off springs in different ways; such as laying eggs, having a live young that can begin
More informationMulticlass and Multi-label Classification
Multiclass and Multi-label Classification INFO-4604, Applied Machine Learning University of Colorado Boulder September 21, 2017 Prof. Michael Paul Today Beyond binary classification All classifiers we
More informationInvestigating Fish Respiration
CHAPTER 31 Fishes and Amphibians Section 31-1 SKILL ACTIVITY Interpreting graphs Investigating Fish Respiration It is well known that a fish dies from lack of oxygen when taken out of water. However, water
More informationInformed search algorithms
Revised by Hankui Zhuo, March 12, 2018 Informed search algorithms Chapter 4 Chapter 4 1 Outline Best-first search A search Heuristics Hill-climbing Simulated annealing Genetic algorithms (briefly) Local
More informationAustralian Journal of Basic and Applied Sciences. Performance Analysis of Different Types of Adder Using 3-Transistor XOR Gate
ISSN:1991-8178 Australian Journal of Basic and Applied Sciences Journal home page: www.ajbasweb.com Performance Analysis of Different Types of Adder Using 3-Transistor XOR Gate Lourdy Nivethitha, V. and
More informationVigilance Behaviour in Barnacle Geese
ASAB Video Practical Vigilance Behaviour in Barnacle Geese Introduction All the barnacle geese (Branta leucopsis) in the world spend the winter in western Europe. Nearly one third of them overwinter in
More informationHow To... Why weigh eggs?
How To... Why weigh eggs? When used in conjunction with body-weight trends, daily egg weight measurements allow accurate management of bird feed allocation during production. Daily egg weight is a sensitive
More informationSupplementary Fig. 1: Comparison of chase parameters for focal pack (a-f, n=1119) and for 4 dogs from 3 other packs (g-m, n=107).
Supplementary Fig. 1: Comparison of chase parameters for focal pack (a-f, n=1119) and for 4 dogs from 3 other packs (g-m, n=107). (a,g) Maximum stride speed, (b,h) maximum tangential acceleration, (c,i)
More informationObjectives. ERTs for the New Beef Industry. Ancient History. The EPD we produce entirely depends on the tools we have to use them.
Bruce Golden, Cal-Poly 6/19/14 Objectives ERTs for the New Beef Industry B. L. Golden 1California Polytechnic State University, San Luis Obispo Ancient History Breeds 1881-1883 First American breed associations
More informationComparative Analysis of Adders Parallel-Prefix Adder for Their Area, Delay and Power Consumption
2018 IJSRST Volume 4 Issue 5 Print ISSN: 2395-6011 Online ISSN: 2395-602X Themed Section: Science and Technology Comparative Analysis of Adders Parallel-Prefix Adder for Their Area, Delay and Power Consumption
More informationEVENTS OR STEPS The events in the story are the steps that the character takes to solve the problem or reach the goal.
LA.K.1.7.1, LA.K.2.1.2 Comprehension Instructional Routine: Identify Story Grammar Preparation/Materials: The Hayloft in the Barn and Otters on the River, two story maps for teacher (use with a transparency
More informationHuman Impact on Sea Turtle Nesting Patterns
Alan Morales Sandoval GIS & GPS APPLICATIONS INTRODUCTION Sea turtles have been around for more than 200 million years. They play an important role in marine ecosystems. Unfortunately, today most species
More informationMulti-Frequency Study of the B3 VLA Sample. I GHz Data
A&A manuscript no. (will be inserted by hand later) Your thesaurus codes are: 13.18.2-11.07.1-11.17.3 ASTRONOMY AND ASTROPHYSICS 3.9.1998 Multi-Frequency Study of the B3 VLA Sample. I. 10.6-GHz Data L.
More informationDesign of 32 bit Parallel Prefix Adders
IOSR Journal of Electronics and Communication Engineering (IOSR-JECE) e-issn: 2278-2834,p- ISSN: 2278-8735. Volume 6, Issue 1 (May. - Jun. 2013), PP 01-06 Design of 32 bit Parallel Prefix Adders P.Chaitanya
More informationKey facts for maximum broiler performance. Changing broiler requires a change of approach
Key facts for maximum broiler performance Changing broiler requires a change of approach Good chick quality = UNIFORMITY everywhere in the supply chain Performance 1. Professional breeder house / management
More informationThe World's Best Jumper
READTHEORY Name Date The World's Best Jumper How high can you jump? If you are like most people, you can probably jump one or two feet high. How high do you think the world's best jumper can jump? A man
More informationThe Heifer Facility Puzzle: The New Puzzle Pieces
The Heifer Facility Puzzle: The New Puzzle Pieces Joe Harner Biological and Agricultural Engineering Kansas State University Manhattan, KS jharner@ksu.edu / 785.532.2900 Dan McFarland Sr Extension Educator
More informationGulf Oil Spill ESSM 651
Gulf Oil Spill ESSM 651 1 Problem statements Introduction The gulf oil spill started on April 20, 2010 when an explosion occurred on the rig, killing 11 workers. The oil spill continued for months until
More informationSHEEP SIRE REFERENCING SCHEMES - NEW OPPORTUNITIES FOR PEDIGREE BREEDERS AND LAMB PRODUCERS a. G. Simm and N.R. Wray
SHEEP SIRE REFERENCING SCHEMES - NEW OPPORTUNITIES FOR PEDIGREE BREEDERS AND LAMB PRODUCERS a G. Simm and N.R. Wray The Scottish Agricultural College Edinburgh, Scotland Summary Sire referencing schemes
More informationIntroduction Methods to Track Changes in Cobra Cobra Project Preferences Cobra New Feature Change Log Summary
Keith Anderson, PMP Implementation Practice Lead, PrimePM EVM Practitioners Forum, July 2014 Introduction Methods to Track Changes in Cobra Cobra Project Preferences Cobra 5.1.2 New Feature Change Log
More informationHOW TO... Feather Sex Day-Old Chicks in the Hatchery
FEATHER SEXING DAY-OLD CHICKS IN THE HATCHERY It is often necessary to sort day-old chicks by sex at the hatchery. To identify females at the parent generation. To separate male and female broilers so
More informationPre-lab Homework Lab 8: Natural Selection
Lab Section: Name: Pre-lab Homework Lab 8: Natural Selection 1. This week's lab uses a mathematical model to simulate the interactions of populations. What is an advantage of using a model like this over
More informationWool Technology and Sheep Breeding
Wool Technology and Sheep Breeding Volume 42, Issue 3 1994 Article 6 Australian Merino central test sire evaluation schemes: operational issues. DJ Cottle JW James Copyright c 1994 Wool Technology and
More information2017 ANIMAL SHELTER STATISTICS
2017 ANIMAL SHELTER STATISTICS INTRODUCTION Dogs and cats are by far Canada s most popular companion animals. In 2017, there were an estimated 7.4 million owned dogs and 9.3 million owned cats living in
More informationNicole Wilde. June 20 & 21, Proudly Presents. Radisson Hotel & Conference Center Ave NW. Edmonton, Alberta
Proudly Presents Nicole Wilde June 20 & 21, 2015 Radisson Hotel & Conference Center 4520 76 Ave NW Edmonton, Alberta Nicole Wilde is the author of ten books including Help for Your Fearful Dog, So You
More informationApproximating the position of a hidden agent in a graph
Approximating the position of a hidden agent in a graph Hannah Guggiari, Alexander Roberts, Alex Scott May 13, 018 Abstract A cat and mouse play a pursuit and evasion game on a connected graph G with n
More informationMexican Gray Wolf Reintroduction
Mexican Gray Wolf Reintroduction New Mexico Supercomputing Challenge Final Report April 2, 2014 Team Number 24 Centennial High School Team Members: Andrew Phillips Teacher: Ms. Hagaman Project Mentor:
More informationEnter Online:
The British Agility Championships The British Novice Stakes October 20 th - 22 nd The College Equestrian Centre Church Road Keysoe,Bedford MK44 2JP Entries Open: 21 st June 2017 Entries Close: 1 st October
More informationLoose Leash Walking. Core Rules Applied:
Loose Leash Walking Many people try to take their dog out for a walk to exercise and at the same time expect them to walk perfectly on leash. Exercise and Loose Leash should be separated into 2 different
More informationSection: 101 (2pm-3pm) 102 (3pm-4pm)
Stat 20 Midterm Exam Instructor: Tessa Childers-Day 12 July 2012 Please write your name and student ID below, and circle your section With your signature, you certify that you have not observed poor or
More informationChapter 18: Categorical data
Chapter 18: Categorical data Self-test answers SELF-TEST Run a multiple regression analysis using Cat Regression.sav with LnObserved as the outcome, and Training, Dance and Interaction as your three predictors.
More informationDesign of Low Power and High Speed Carry Select Adder Using Brent Kung Adder
Design of Low Power and High Speed Carry Select Adder Using Brent Kung Adder Dr.K.Srinivasulu Professor, Dept of ECE, Malla Reddy Collage of Engineering. Abstract: The binary addition is the basic arithmetic
More informationRumination Monitoring White Paper
Rumination Monitoring White Paper Introduction to Rumination Monitoring Summary Rumination is a proven direct indicator of cow wellbeing and health. Dairy producers, veterinarians and nutritionists have
More informationData Mining Classification: Alternative Techniques. Lecture Notes for Chapter 4. Rule-Based. Introduction to Data Mining, 2 nd Edition
Data Mining Classification: Alternative Techniques Lecture Notes for Chapter 4 RuleBased Introduction to Data Mining, 2 nd Edition by Tan, Steinbach, Karpatne, Kumar RuleBased Classifier Classify records
More informationActivities. Life in the Arctic Tundra. Grades: PreK K, 1 2, 3 5, 6 8
Activities Life in the Arctic Tundra Grades: PreK K, 1 2, 3 5, 6 8 Overview A series of activities reinforce the learning content: padded gloves insulate like fur, folk tales give new meaning to the long
More informationCourse: Canine Massage and Bodywork Certification Course Part A Cranial Trunk and Thoracic Appendicular System. Movers of the Forelimb, Neck, and Head
Course: Canine Massage and Bodywork Certification Course Part A Cranial Trunk and Thoracic Appendicular System. Movers of the Forelimb, Neck, and Head Course Number: CN4000A Course Instructors: Beverly
More informationTitle. Grade level. Time. Student Target. PART 3 Lesson: Populations. PART 3 Activity: Turtles, Turtle Everywhere! minutes
Title PART 3 Lesson: Populations PART 3 Activity: Turtles, Turtle Everywhere! Grade level 3-5 Time 60 minutes Student Target SC.3.N.1.1 Raise questions about the natural world, investigate them individually
More informationFraction Approximation: Closer to Zero, One-half or One whole? CCSS: 3.NF.3, 4.NF.2 VA SOLs: 3.3, 4.2, 5.2
Fraction Approximation: Closer to Zero, One-half or One whole? CCSS: 3.NF.3, 4.NF.2 VA SOLs: 3.3, 4.2, 5.2 Fraction Approximation: Closer to zero, one-half, or one whole? Have students decide whether their
More informationAccommodation Process for Comfort Animal in Campus Housing and Responsibilities of the Comfort Animal Owner
Accommodation Process for Comfort Animal in Campus Housing and Responsibilities of the Comfort Animal Owner According to UNT Policy 16.002, requests for accommodation for a comfort animal must be filed
More informationRelationship Between Eye Color and Success in Anatomy. Sam Holladay IB Math Studies Mr. Saputo 4/3/15
Relationship Between Eye Color and Success in Anatomy Sam Holladay IB Math Studies Mr. Saputo 4/3/15 Table of Contents Section A: Introduction.. 2 Section B: Information/Measurement... 3 Section C: Mathematical
More informationTailoring a terminal sire breeding program for the west
Tailoring a terminal sire breeding program for the west Ron Lewis, Department of Animal Science, University of Nebraska-Lincoln Utah Wool Growers Association Leading Edge Sheep Production Part II Little
More informationThe Kaggle Competitions: An Introduction to CAMCOS Fall 2015
The Kaggle Competitions: An Introduction to CAMCOS Fall 15 Guangliang Chen Math/Stats Colloquium San Jose State University August 6, 15 Outline Introduction to Kaggle Description of projects Summary Guangliang
More informationNordic Cattle Genetic Evaluation a tool for practical breeding with red breeds
Nordic Cattle Genetic Evaluation a tool for practical breeding with red breeds Gert Pedersen Aamand, Nordic Cattle Genetic Evaluation, Udkaersvej 15, DK-8200 Aarhus N, Denmark e-mail: gap@landscentret.dk
More informationA Genetic Comparison of Standard and Miniature Poodles based on autosomal markers and DLA class II haplotypes.
A Genetic Comparison of Standard and Miniature Poodles based on autosomal markers and DLA class II haplotypes. Niels C. Pedersen, 1 Lorna J. Kennedy 2 1 Center for Companion Animal Health, School of Veterinary
More informationGenetic approaches to improving lamb survival
Genetic approaches to improving lamb survival SBRT, Nottingham - 18-nov-2017 Mark Young CIEL United Kingdom Forbes Brien University of Adelaide Australia UK Agri-Tech Centres Motivated by vision to drive
More informationProblems from The Calculus of Friendship:
Problems from The Calculus of Friendship: Worth Corresponding About Carmel Schettino Inspired by Rick Parris & Ron Lancaster A Wonderful Narrative Book of relationship Read it in '08 gave as gift NYT Opinionator
More informationOIE standards : procedures, model certificates
OIE standards : procedures, model certificates Workshop for OIE National Focal Points for Animal Production Food Safety Singapore, October 2010 Thanks to: Caroline Planté / Wim Pilgrim Alexandre Bouchot
More informationActive sensing. Ehud Ahissar
Active sensing Ehud Ahissar 1 Active sensing Passive vs active sensing (touch) Comparison across senses Basic coding principles -------- Perceptual loops Sensation-targeted motor control Proprioception
More informationIndigo Sapphire Bear. Newfoundland. Indigo Sapphire Bear. January. Dog's name: DR. NEALE FRETWELL. R&D Director
Indigo Sapphire Bear Dog's name: Indigo Sapphire Bear This certifies the authenticity of Indigo Sapphire Bear's canine genetic background as determined following careful analysis of more than 300 genetic
More informationComparative Evaluation of Online and Paper & Pencil Forms for the Iowa Assessments ITP Research Series
Comparative Evaluation of Online and Paper & Pencil Forms for the Iowa Assessments ITP Research Series Catherine J. Welch Stephen B. Dunbar Heather Rickels Keyu Chen ITP Research Series 2014.2 A Comparative
More informationC R H G E K. 1 Solve the puzzle. lion. parrot. crocodile. flamingo. snake. tortoise. horse. zebra. elephant. eagle duck. monkey. Classify the animals.
Wild animals 1 Solve the puzzle. Z E C R lion crocodile snake parrot H G 0 E L 0 E K flamingo tortoise D horse zebra elephant Classify the animals. monkey eagle duck Mammals Birds Reptiles 40 Unit 4 Wild
More informationThe EVM + AGILE Anthology
The EVM + AGILE Anthology Ray W. Stratton, PMP, EVP Management Technologies raystratton@mgmt-technologies.com Project Management Institute California Inland Empire Chapter 17 February 2015 Who is this
More informationDunbia 2017 Dunbia 2017
Dunbia 2017 2017 Thinking differently about collecting data 1) Overview of SPiLAMM project 2) Technology developments 3) Analysis and farmer feedback 4) Drivers and barriers to new technologies 5) Using
More informationThe Dog of Pompeii. historical fiction by Louis Untermeyer
The Dog of Pompeii historical fiction by Louis Untermeyer Directions 1. Preview the questions you ll be asked first. Make notes of them on your CBT worksheet. 2. Read the selection, The Dog of Pompeii.
More informationPROGRESS REPORT for COOPERATIVE BOBCAT RESEARCH PROJECT. Period Covered: 1 April 30 June Prepared by
PROGRESS REPORT for COOPERATIVE BOBCAT RESEARCH PROJECT Period Covered: 1 April 30 June 2014 Prepared by John A. Litvaitis, Tyler Mahard, Rory Carroll, and Marian K. Litvaitis Department of Natural Resources
More informationPerplexity of n-gram and dependency language models
Perplexity of n-gram and dependency language models Martin Popel, David Mareček ÚFAL, Charles University in Prague TSD, 13th International Conference on Text, Speech and Dialogue September 8, 2010, Brno
More informationComparison of Parallel Prefix Adders Performance in an FPGA
International Journal of Engineering Research and Development e-issn: 2278-067X, p-issn: 2278-800X, www.ijerd.com Volume 3, Issue 6 (September 2012), PP. 62-67 Comparison of Parallel Prefix Adders Performance
More informationSTATISTICAL REPORT. Preliminary Analysis of the Second Collaborative Study of the Hard Surface Carrier Test
STATISTICAL REPORT To: From: Subject: Diane Boesenberg, Reckitt Benckiser Emily Mitchell, Product Science Branch, Antimicrobials Division/Office of Pesticide Programs/US EPA Martin Hamilton, Statistician
More informationIt Is Raining Cats. Margaret Kwok St #: Biology 438
It Is Raining Cats Margaret Kwok St #: 80445992 Biology 438 Abstract Cats are known to right themselves by rotating their bodies while falling through the air and despite being released from almost any
More informationLABORATORY EXERCISE 6: CLADISTICS I
Biology 4415/5415 Evolution LABORATORY EXERCISE 6: CLADISTICS I Take a group of organisms. Let s use five: a lungfish, a frog, a crocodile, a flamingo, and a human. How to reconstruct their relationships?
More information