SlideShare a Scribd company logo
The Allen AI Science
Challenge &
DeepHack.Q&A
St. Petersburg Data Science Meetup #6, Feb 19th, 2016
Q: When athletes begin to exercise, their heart rates and respiration rates
increase. At what level of organization does the human body coordinate these
functions?
A. at the tissue level
B. at the organ level
C. at the system level
D. at the cellular level
Wed 7 Oct 2015 – Sat 13 Feb 2016
Stage 1: 800 teams (>1000 participants),
Stage 2: 170 teams
https://www.kaggle.com/c/the-allen-ai-science-challenge
2700 questions - train set
8132 questions - validation set
21298 questions - final test set
DeepHack Q&A qa.deephack.me/
Qualification round: Top-50 participants with the
highest scores
Rough competition: Kaggle Top-40 to get to the Top-50 o_O
Winter ML school + hackathon: 31st, Jan - 5th Feb, 2016
GP team created at Jan, 31st from the four teams
The final 30 minutes of the hackathon: https:
//www.youtube.com/watch?v=tCKL5vbiHuo
Pavel Kalaidin (VK)
Marat Zainutdinov (Quantbrothers)
Roman Trusov (ITMO University)
Artyom Korkhov (Zvooq)
Igor Shilov (Zvooq)
Timur Luguev (Clevapi)
Ilyas Luguev (Clevapi)
Team Generation Gap
DeepHack: 1st, ~0.556
Allen AI: 7th, 0.55059
Datasets
ck12.org
wikipedia.org (science subset)
flashcards: studystack.com,
quizlet.com
Topic at the forum: https://www.kaggle.com/c/the-allen-ai-science-challenge/forums/t/16877/external-data-
repository
Hail to Lucene
Lucene
Question
a) ans1
b) ans2
c) ans3
d) ans4
Question ans1
Question ans4
Question ans3
Question ans2
0.5
0.4
0.02
0.01
...
0.5
0.4
0.02
0.01
...
0.5
0.4
0.02
0.01
...
0.5
0.4
0.02
0.01
...
Wiki ck12
quizlets
Stemming,stopwords
Custom queries rule
Lucene scores: https://lucene.apache.
org/core/3_5_0/api/core/org/apache/lucene/search/Similarity.html
AdaGram (a.k.a Reptil)
Breaking Sticks and Ambiguities with Adaptive Skip-gram: http:
//arxiv.org/abs/1502.07257
Reference implementation in Julia: https://github.
com/sbos/AdaGram.jl
reptil art cultur final play
signific role folklor
religion popular cultur moch
peopl noun coldblood anim
scale general move stomach
short leg exampl snake lizard
turtl noun aw person
Model trained like this: sh train.sh --min-freq 20 --window 5 --workers 40 --epochs 5
--dim 300 --alpha 0.1 corpus.txt adam.dict adam.model
Number of prototypes is 5 by default.
AdaGram (a.k.a Reptil) approach
[0.42, 0.55, 0.08, …]
N-grams PMI
x, y - Ngrams
Example 1-gram -> 1-gram
unit -> state
magnet -> field
carbon -> dioxid
million -> year
year -> ago
amino -> acid
Example 1-gram -> 3-gram
around -> million year ago
period -> million year ago
forc -> van der waal
fossil -> million year ago
nobel -> prize physiolog medicin
date -> million year ago
mercuri -> venus earth mar
N-grams PMI
greatest contributor air pollut unit state
What is the greatest contributor to air pollution
in the United States?
greatest
contributor
air
...
greatest contributor
contributor air
air pollut
...
1-grams 2-grams 3-grams
Power plants
power plant
power
plant
1-grams
power plant
2-grams
...
Scores
Fail Story
TL;DR wasted tons of time, got ~0.3 in
almost all approaches
LSA + Lucene
Corpus
LSA
TI_1
TI_2
TI_n
Lucene
qa pair 1
qa pair 2
qa pair 3
qa pair 4
Queries in
topic
indices
Result: for each qa pair,
max(s1...sn)
Gave 1% improvement over
basic Lucene; but took
EXTREMELY long time to
process :(
Syntax co-occurrence
nobel chemistry prize 517
national science academy 445
long time period 340
also role play 306
nobel physic prize 279
national medical library 273
carbon water dioxide 261
second thermodynamics law 247
speed sound of_pobj
density population compound
take place dobj
link external compound 0.3 :(
word2vec combinations
Wanted to capture the intersection of meanings, but didn’t know
how to combine word2vec representations
TFIDFqa pairs
Combinations of
question tokens
Combinations of
answer tokens
Cosine
similarity
Max score ~ 0.3 :( even with careful kw filtering
word2gauss didn’t help too
Averaging Neural networks (1st encounter)
w2v_dim = 300
vec_q = mean(w2v(Q))
vec_c = mean(w2v(Ac))
vec_w = mean(w2v(Aw))
cos_sim(vec_q, vec_c) > cos_sim(vec_q, vec_w)
Averaging Neural networks (1st encounter)
w2v_dim = 300
vec_q = mean(w2v(Q))
vec_c = mean(w2v(Ac))
vec_w = mean(w2v(Aw))
a = CNN(w2v(X))
vec_x = mean(w2v(X) * a)
cos_sim(vec_q, vec_c) > cos_sim(vec_q, vec_w)
Semantic Neural networks (2nd encounter)
+ Paragraphs
LSTM = LSTM(w2v)
LSTM(s1 | s2) > LSTM(s1 | s3) if s1 and s2 are from the same
paragraph, while s1 and s3 are not
LSTM(a, b) is low then a and b are from the same paragraph (energy
based learning)
Loss = max(0, M - LSTM(s1, s2) + LSTM(s1, s3))
Score: 0.26
Siamese architecture
Hinge Loss
Margin
Reading Neural networks (3rd encounter)
+ Lots of paragraphs
+ Search Engine
+ A survey:
- bigrams are not accounted
- main idea (keywords) of a sentence is not recognized
Reading Neural networks (3rd encounter)
+ Lots of paragraphs
+ Search Engine
+ A survey:
- bigrams are not accounted
- main idea (keywords) of a sentence is not recognized
Reading Neural networks (3rd encounter)
All we want is to know if a sentence is from a paragraph to be
able to rerank lucene scores.
Hinge Loss
Margin
LSTM(P)
LSTM(s1)
LSTM(s2)
Reading Neural networks (3rd encounter)
sentences -> LSTM -> Dense NN -> Embedding
w2v -> LSTM -> Dense NN -> Embedding
w2v -> Mean -> Dense NN -> Embedding
Neural networks. Learned lessons.
Start as small as possible
Corruption is important for siamese networks
Learning curve is misleading in NLP
Lessons learned
Start early - wasted two first months of the competition (but had a week
of 24/7 hackathon at the end)
No stickers in the team channel (except with Yann LeCun on a good submit)
Common toolbox is nice
A dedicated server is a good thing to have (no need in AWS spot instances)
Experiment fast, fail early
Team work means a lot

More Related Content

Similar to The Allen AI Science Challenge

Introduction to Algorithms
Introduction to AlgorithmsIntroduction to Algorithms
Introduction to Algorithms
Venkatesh Iyer
 
2.03.Asymptotic_analysis.pptx
2.03.Asymptotic_analysis.pptx2.03.Asymptotic_analysis.pptx
2.03.Asymptotic_analysis.pptx
ssuser1fb3df
 
String kmp
String kmpString kmp
String kmp
thinkphp
 
Computing k-rank Answers with Ontological CP-nets
Computing k-rank Answers with Ontological CP-netsComputing k-rank Answers with Ontological CP-nets
Computing k-rank Answers with Ontological CP-nets
Oana Tifrea-Marciuska
 
A Network-Aware Approach for Searching As-You-Type in Social Media
A Network-Aware Approach for Searching As-You-Type in Social MediaA Network-Aware Approach for Searching As-You-Type in Social Media
A Network-Aware Approach for Searching As-You-Type in Social Media
INRIA-OAK
 
Computer Science Engineering : Data structure & algorithm, THE GATE ACADEMY
Computer Science Engineering : Data structure & algorithm, THE GATE ACADEMYComputer Science Engineering : Data structure & algorithm, THE GATE ACADEMY
Computer Science Engineering : Data structure & algorithm, THE GATE ACADEMY
klirantga
 
Machine Learning
Machine LearningMachine Learning
Machine Learningbutest
 
Lecture 2 data structures & algorithms - sorting techniques
Lecture 2  data structures & algorithms - sorting techniquesLecture 2  data structures & algorithms - sorting techniques
Lecture 2 data structures & algorithms - sorting techniquesDharmendra Prasad
 
Software-Praktikum SoSe 2005 Lehrstuhl fuer Maschinelles ...
Software-Praktikum SoSe 2005 Lehrstuhl fuer Maschinelles ...Software-Praktikum SoSe 2005 Lehrstuhl fuer Maschinelles ...
Software-Praktikum SoSe 2005 Lehrstuhl fuer Maschinelles ...butest
 
Stack squeues lists
Stack squeues listsStack squeues lists
Stack squeues lists
James Wong
 
Stacks queues lists
Stacks queues listsStacks queues lists
Stacks queues lists
Harry Potter
 
Stacks queues lists
Stacks queues listsStacks queues lists
Stacks queues lists
Luis Goldster
 
Stacksqueueslists
StacksqueueslistsStacksqueueslists
Stacksqueueslists
Fraboni Ec
 
Stacks queues lists
Stacks queues listsStacks queues lists
Stacks queues lists
Young Alista
 
Stacks queues lists
Stacks queues listsStacks queues lists
Stacks queues lists
Tony Nguyen
 
algo1-aa-gist_typed.pdf
algo1-aa-gist_typed.pdfalgo1-aa-gist_typed.pdf
algo1-aa-gist_typed.pdf
prajjwalbajpai1
 
time_complexity_list_02_04_2024_22_pages.pdf
time_complexity_list_02_04_2024_22_pages.pdftime_complexity_list_02_04_2024_22_pages.pdf
time_complexity_list_02_04_2024_22_pages.pdf
SrinivasaReddyPolamR
 
Introduction to computing Processing and performance.pdf
Introduction to computing Processing and performance.pdfIntroduction to computing Processing and performance.pdf
Introduction to computing Processing and performance.pdf
TulasiramKandula1
 
introduction to algorithm for beginneer1
introduction to algorithm for beginneer1introduction to algorithm for beginneer1
introduction to algorithm for beginneer1
ranjankumarbehera14
 
Lecture38
Lecture38Lecture38
Lecture38
David Evans
 

Similar to The Allen AI Science Challenge (20)

Introduction to Algorithms
Introduction to AlgorithmsIntroduction to Algorithms
Introduction to Algorithms
 
2.03.Asymptotic_analysis.pptx
2.03.Asymptotic_analysis.pptx2.03.Asymptotic_analysis.pptx
2.03.Asymptotic_analysis.pptx
 
String kmp
String kmpString kmp
String kmp
 
Computing k-rank Answers with Ontological CP-nets
Computing k-rank Answers with Ontological CP-netsComputing k-rank Answers with Ontological CP-nets
Computing k-rank Answers with Ontological CP-nets
 
A Network-Aware Approach for Searching As-You-Type in Social Media
A Network-Aware Approach for Searching As-You-Type in Social MediaA Network-Aware Approach for Searching As-You-Type in Social Media
A Network-Aware Approach for Searching As-You-Type in Social Media
 
Computer Science Engineering : Data structure & algorithm, THE GATE ACADEMY
Computer Science Engineering : Data structure & algorithm, THE GATE ACADEMYComputer Science Engineering : Data structure & algorithm, THE GATE ACADEMY
Computer Science Engineering : Data structure & algorithm, THE GATE ACADEMY
 
Machine Learning
Machine LearningMachine Learning
Machine Learning
 
Lecture 2 data structures & algorithms - sorting techniques
Lecture 2  data structures & algorithms - sorting techniquesLecture 2  data structures & algorithms - sorting techniques
Lecture 2 data structures & algorithms - sorting techniques
 
Software-Praktikum SoSe 2005 Lehrstuhl fuer Maschinelles ...
Software-Praktikum SoSe 2005 Lehrstuhl fuer Maschinelles ...Software-Praktikum SoSe 2005 Lehrstuhl fuer Maschinelles ...
Software-Praktikum SoSe 2005 Lehrstuhl fuer Maschinelles ...
 
Stack squeues lists
Stack squeues listsStack squeues lists
Stack squeues lists
 
Stacks queues lists
Stacks queues listsStacks queues lists
Stacks queues lists
 
Stacks queues lists
Stacks queues listsStacks queues lists
Stacks queues lists
 
Stacksqueueslists
StacksqueueslistsStacksqueueslists
Stacksqueueslists
 
Stacks queues lists
Stacks queues listsStacks queues lists
Stacks queues lists
 
Stacks queues lists
Stacks queues listsStacks queues lists
Stacks queues lists
 
algo1-aa-gist_typed.pdf
algo1-aa-gist_typed.pdfalgo1-aa-gist_typed.pdf
algo1-aa-gist_typed.pdf
 
time_complexity_list_02_04_2024_22_pages.pdf
time_complexity_list_02_04_2024_22_pages.pdftime_complexity_list_02_04_2024_22_pages.pdf
time_complexity_list_02_04_2024_22_pages.pdf
 
Introduction to computing Processing and performance.pdf
Introduction to computing Processing and performance.pdfIntroduction to computing Processing and performance.pdf
Introduction to computing Processing and performance.pdf
 
introduction to algorithm for beginneer1
introduction to algorithm for beginneer1introduction to algorithm for beginneer1
introduction to algorithm for beginneer1
 
Lecture38
Lecture38Lecture38
Lecture38
 

Recently uploaded

Malana- Gimlet Market Analysis (Portfolio 2)
Malana- Gimlet Market Analysis (Portfolio 2)Malana- Gimlet Market Analysis (Portfolio 2)
Malana- Gimlet Market Analysis (Portfolio 2)
TravisMalana
 
一比一原版(Dalhousie毕业证书)达尔豪斯大学毕业证如何办理
一比一原版(Dalhousie毕业证书)达尔豪斯大学毕业证如何办理一比一原版(Dalhousie毕业证书)达尔豪斯大学毕业证如何办理
一比一原版(Dalhousie毕业证书)达尔豪斯大学毕业证如何办理
mzpolocfi
 
做(mqu毕业证书)麦考瑞大学毕业证硕士文凭证书学费发票原版一模一样
做(mqu毕业证书)麦考瑞大学毕业证硕士文凭证书学费发票原版一模一样做(mqu毕业证书)麦考瑞大学毕业证硕士文凭证书学费发票原版一模一样
做(mqu毕业证书)麦考瑞大学毕业证硕士文凭证书学费发票原版一模一样
axoqas
 
Ch03-Managing the Object-Oriented Information Systems Project a.pdf
Ch03-Managing the Object-Oriented Information Systems Project a.pdfCh03-Managing the Object-Oriented Information Systems Project a.pdf
Ch03-Managing the Object-Oriented Information Systems Project a.pdf
haila53
 
STATATHON: Unleashing the Power of Statistics in a 48-Hour Knowledge Extravag...
STATATHON: Unleashing the Power of Statistics in a 48-Hour Knowledge Extravag...STATATHON: Unleashing the Power of Statistics in a 48-Hour Knowledge Extravag...
STATATHON: Unleashing the Power of Statistics in a 48-Hour Knowledge Extravag...
sameer shah
 
一比一原版(CBU毕业证)卡普顿大学毕业证如何办理
一比一原版(CBU毕业证)卡普顿大学毕业证如何办理一比一原版(CBU毕业证)卡普顿大学毕业证如何办理
一比一原版(CBU毕业证)卡普顿大学毕业证如何办理
ahzuo
 
Adjusting OpenMP PageRank : SHORT REPORT / NOTES
Adjusting OpenMP PageRank : SHORT REPORT / NOTESAdjusting OpenMP PageRank : SHORT REPORT / NOTES
Adjusting OpenMP PageRank : SHORT REPORT / NOTES
Subhajit Sahu
 
一比一原版(爱大毕业证书)爱丁堡大学毕业证如何办理
一比一原版(爱大毕业证书)爱丁堡大学毕业证如何办理一比一原版(爱大毕业证书)爱丁堡大学毕业证如何办理
一比一原版(爱大毕业证书)爱丁堡大学毕业证如何办理
g4dpvqap0
 
Data_and_Analytics_Essentials_Architect_an_Analytics_Platform.pptx
Data_and_Analytics_Essentials_Architect_an_Analytics_Platform.pptxData_and_Analytics_Essentials_Architect_an_Analytics_Platform.pptx
Data_and_Analytics_Essentials_Architect_an_Analytics_Platform.pptx
AnirbanRoy608946
 
哪里卖(usq毕业证书)南昆士兰大学毕业证研究生文凭证书托福证书原版一模一样
哪里卖(usq毕业证书)南昆士兰大学毕业证研究生文凭证书托福证书原版一模一样哪里卖(usq毕业证书)南昆士兰大学毕业证研究生文凭证书托福证书原版一模一样
哪里卖(usq毕业证书)南昆士兰大学毕业证研究生文凭证书托福证书原版一模一样
axoqas
 
一比一原版(Bradford毕业证书)布拉德福德大学毕业证如何办理
一比一原版(Bradford毕业证书)布拉德福德大学毕业证如何办理一比一原版(Bradford毕业证书)布拉德福德大学毕业证如何办理
一比一原版(Bradford毕业证书)布拉德福德大学毕业证如何办理
mbawufebxi
 
ViewShift: Hassle-free Dynamic Policy Enforcement for Every Data Lake
ViewShift: Hassle-free Dynamic Policy Enforcement for Every Data LakeViewShift: Hassle-free Dynamic Policy Enforcement for Every Data Lake
ViewShift: Hassle-free Dynamic Policy Enforcement for Every Data Lake
Walaa Eldin Moustafa
 
一比一原版(UniSA毕业证书)南澳大学毕业证如何办理
一比一原版(UniSA毕业证书)南澳大学毕业证如何办理一比一原版(UniSA毕业证书)南澳大学毕业证如何办理
一比一原版(UniSA毕业证书)南澳大学毕业证如何办理
slg6lamcq
 
Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...
Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...
Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...
John Andrews
 
06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Dat...
06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Dat...06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Dat...
06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Dat...
Timothy Spann
 
Algorithmic optimizations for Dynamic Levelwise PageRank (from STICD) : SHORT...
Algorithmic optimizations for Dynamic Levelwise PageRank (from STICD) : SHORT...Algorithmic optimizations for Dynamic Levelwise PageRank (from STICD) : SHORT...
Algorithmic optimizations for Dynamic Levelwise PageRank (from STICD) : SHORT...
Subhajit Sahu
 
办(uts毕业证书)悉尼科技大学毕业证学历证书原版一模一样
办(uts毕业证书)悉尼科技大学毕业证学历证书原版一模一样办(uts毕业证书)悉尼科技大学毕业证学历证书原版一模一样
办(uts毕业证书)悉尼科技大学毕业证学历证书原版一模一样
apvysm8
 
原版制作(Deakin毕业证书)迪肯大学毕业证学位证一模一样
原版制作(Deakin毕业证书)迪肯大学毕业证学位证一模一样原版制作(Deakin毕业证书)迪肯大学毕业证学位证一模一样
原版制作(Deakin毕业证书)迪肯大学毕业证学位证一模一样
u86oixdj
 
一比一原版(Adelaide毕业证书)阿德莱德大学毕业证如何办理
一比一原版(Adelaide毕业证书)阿德莱德大学毕业证如何办理一比一原版(Adelaide毕业证书)阿德莱德大学毕业证如何办理
一比一原版(Adelaide毕业证书)阿德莱德大学毕业证如何办理
slg6lamcq
 
Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...
Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...
Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...
Subhajit Sahu
 

Recently uploaded (20)

Malana- Gimlet Market Analysis (Portfolio 2)
Malana- Gimlet Market Analysis (Portfolio 2)Malana- Gimlet Market Analysis (Portfolio 2)
Malana- Gimlet Market Analysis (Portfolio 2)
 
一比一原版(Dalhousie毕业证书)达尔豪斯大学毕业证如何办理
一比一原版(Dalhousie毕业证书)达尔豪斯大学毕业证如何办理一比一原版(Dalhousie毕业证书)达尔豪斯大学毕业证如何办理
一比一原版(Dalhousie毕业证书)达尔豪斯大学毕业证如何办理
 
做(mqu毕业证书)麦考瑞大学毕业证硕士文凭证书学费发票原版一模一样
做(mqu毕业证书)麦考瑞大学毕业证硕士文凭证书学费发票原版一模一样做(mqu毕业证书)麦考瑞大学毕业证硕士文凭证书学费发票原版一模一样
做(mqu毕业证书)麦考瑞大学毕业证硕士文凭证书学费发票原版一模一样
 
Ch03-Managing the Object-Oriented Information Systems Project a.pdf
Ch03-Managing the Object-Oriented Information Systems Project a.pdfCh03-Managing the Object-Oriented Information Systems Project a.pdf
Ch03-Managing the Object-Oriented Information Systems Project a.pdf
 
STATATHON: Unleashing the Power of Statistics in a 48-Hour Knowledge Extravag...
STATATHON: Unleashing the Power of Statistics in a 48-Hour Knowledge Extravag...STATATHON: Unleashing the Power of Statistics in a 48-Hour Knowledge Extravag...
STATATHON: Unleashing the Power of Statistics in a 48-Hour Knowledge Extravag...
 
一比一原版(CBU毕业证)卡普顿大学毕业证如何办理
一比一原版(CBU毕业证)卡普顿大学毕业证如何办理一比一原版(CBU毕业证)卡普顿大学毕业证如何办理
一比一原版(CBU毕业证)卡普顿大学毕业证如何办理
 
Adjusting OpenMP PageRank : SHORT REPORT / NOTES
Adjusting OpenMP PageRank : SHORT REPORT / NOTESAdjusting OpenMP PageRank : SHORT REPORT / NOTES
Adjusting OpenMP PageRank : SHORT REPORT / NOTES
 
一比一原版(爱大毕业证书)爱丁堡大学毕业证如何办理
一比一原版(爱大毕业证书)爱丁堡大学毕业证如何办理一比一原版(爱大毕业证书)爱丁堡大学毕业证如何办理
一比一原版(爱大毕业证书)爱丁堡大学毕业证如何办理
 
Data_and_Analytics_Essentials_Architect_an_Analytics_Platform.pptx
Data_and_Analytics_Essentials_Architect_an_Analytics_Platform.pptxData_and_Analytics_Essentials_Architect_an_Analytics_Platform.pptx
Data_and_Analytics_Essentials_Architect_an_Analytics_Platform.pptx
 
哪里卖(usq毕业证书)南昆士兰大学毕业证研究生文凭证书托福证书原版一模一样
哪里卖(usq毕业证书)南昆士兰大学毕业证研究生文凭证书托福证书原版一模一样哪里卖(usq毕业证书)南昆士兰大学毕业证研究生文凭证书托福证书原版一模一样
哪里卖(usq毕业证书)南昆士兰大学毕业证研究生文凭证书托福证书原版一模一样
 
一比一原版(Bradford毕业证书)布拉德福德大学毕业证如何办理
一比一原版(Bradford毕业证书)布拉德福德大学毕业证如何办理一比一原版(Bradford毕业证书)布拉德福德大学毕业证如何办理
一比一原版(Bradford毕业证书)布拉德福德大学毕业证如何办理
 
ViewShift: Hassle-free Dynamic Policy Enforcement for Every Data Lake
ViewShift: Hassle-free Dynamic Policy Enforcement for Every Data LakeViewShift: Hassle-free Dynamic Policy Enforcement for Every Data Lake
ViewShift: Hassle-free Dynamic Policy Enforcement for Every Data Lake
 
一比一原版(UniSA毕业证书)南澳大学毕业证如何办理
一比一原版(UniSA毕业证书)南澳大学毕业证如何办理一比一原版(UniSA毕业证书)南澳大学毕业证如何办理
一比一原版(UniSA毕业证书)南澳大学毕业证如何办理
 
Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...
Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...
Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...
 
06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Dat...
06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Dat...06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Dat...
06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Dat...
 
Algorithmic optimizations for Dynamic Levelwise PageRank (from STICD) : SHORT...
Algorithmic optimizations for Dynamic Levelwise PageRank (from STICD) : SHORT...Algorithmic optimizations for Dynamic Levelwise PageRank (from STICD) : SHORT...
Algorithmic optimizations for Dynamic Levelwise PageRank (from STICD) : SHORT...
 
办(uts毕业证书)悉尼科技大学毕业证学历证书原版一模一样
办(uts毕业证书)悉尼科技大学毕业证学历证书原版一模一样办(uts毕业证书)悉尼科技大学毕业证学历证书原版一模一样
办(uts毕业证书)悉尼科技大学毕业证学历证书原版一模一样
 
原版制作(Deakin毕业证书)迪肯大学毕业证学位证一模一样
原版制作(Deakin毕业证书)迪肯大学毕业证学位证一模一样原版制作(Deakin毕业证书)迪肯大学毕业证学位证一模一样
原版制作(Deakin毕业证书)迪肯大学毕业证学位证一模一样
 
一比一原版(Adelaide毕业证书)阿德莱德大学毕业证如何办理
一比一原版(Adelaide毕业证书)阿德莱德大学毕业证如何办理一比一原版(Adelaide毕业证书)阿德莱德大学毕业证如何办理
一比一原版(Adelaide毕业证书)阿德莱德大学毕业证如何办理
 
Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...
Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...
Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...
 

The Allen AI Science Challenge

  • 1. The Allen AI Science Challenge & DeepHack.Q&A St. Petersburg Data Science Meetup #6, Feb 19th, 2016
  • 2. Q: When athletes begin to exercise, their heart rates and respiration rates increase. At what level of organization does the human body coordinate these functions? A. at the tissue level B. at the organ level C. at the system level D. at the cellular level Wed 7 Oct 2015 – Sat 13 Feb 2016 Stage 1: 800 teams (>1000 participants), Stage 2: 170 teams https://www.kaggle.com/c/the-allen-ai-science-challenge 2700 questions - train set 8132 questions - validation set 21298 questions - final test set
  • 3. DeepHack Q&A qa.deephack.me/ Qualification round: Top-50 participants with the highest scores Rough competition: Kaggle Top-40 to get to the Top-50 o_O Winter ML school + hackathon: 31st, Jan - 5th Feb, 2016 GP team created at Jan, 31st from the four teams The final 30 minutes of the hackathon: https: //www.youtube.com/watch?v=tCKL5vbiHuo
  • 4. Pavel Kalaidin (VK) Marat Zainutdinov (Quantbrothers) Roman Trusov (ITMO University) Artyom Korkhov (Zvooq) Igor Shilov (Zvooq) Timur Luguev (Clevapi) Ilyas Luguev (Clevapi) Team Generation Gap DeepHack: 1st, ~0.556 Allen AI: 7th, 0.55059
  • 5. Datasets ck12.org wikipedia.org (science subset) flashcards: studystack.com, quizlet.com Topic at the forum: https://www.kaggle.com/c/the-allen-ai-science-challenge/forums/t/16877/external-data- repository
  • 6. Hail to Lucene Lucene Question a) ans1 b) ans2 c) ans3 d) ans4 Question ans1 Question ans4 Question ans3 Question ans2 0.5 0.4 0.02 0.01 ... 0.5 0.4 0.02 0.01 ... 0.5 0.4 0.02 0.01 ... 0.5 0.4 0.02 0.01 ... Wiki ck12 quizlets Stemming,stopwords
  • 7. Custom queries rule Lucene scores: https://lucene.apache. org/core/3_5_0/api/core/org/apache/lucene/search/Similarity.html
  • 8. AdaGram (a.k.a Reptil) Breaking Sticks and Ambiguities with Adaptive Skip-gram: http: //arxiv.org/abs/1502.07257 Reference implementation in Julia: https://github. com/sbos/AdaGram.jl
  • 9. reptil art cultur final play signific role folklor religion popular cultur moch peopl noun coldblood anim scale general move stomach short leg exampl snake lizard turtl noun aw person
  • 10. Model trained like this: sh train.sh --min-freq 20 --window 5 --workers 40 --epochs 5 --dim 300 --alpha 0.1 corpus.txt adam.dict adam.model Number of prototypes is 5 by default.
  • 11. AdaGram (a.k.a Reptil) approach [0.42, 0.55, 0.08, …]
  • 12. N-grams PMI x, y - Ngrams Example 1-gram -> 1-gram unit -> state magnet -> field carbon -> dioxid million -> year year -> ago amino -> acid Example 1-gram -> 3-gram around -> million year ago period -> million year ago forc -> van der waal fossil -> million year ago nobel -> prize physiolog medicin date -> million year ago mercuri -> venus earth mar
  • 13. N-grams PMI greatest contributor air pollut unit state What is the greatest contributor to air pollution in the United States? greatest contributor air ... greatest contributor contributor air air pollut ... 1-grams 2-grams 3-grams Power plants power plant power plant 1-grams power plant 2-grams ...
  • 15. Fail Story TL;DR wasted tons of time, got ~0.3 in almost all approaches
  • 16. LSA + Lucene Corpus LSA TI_1 TI_2 TI_n Lucene qa pair 1 qa pair 2 qa pair 3 qa pair 4 Queries in topic indices Result: for each qa pair, max(s1...sn) Gave 1% improvement over basic Lucene; but took EXTREMELY long time to process :(
  • 17. Syntax co-occurrence nobel chemistry prize 517 national science academy 445 long time period 340 also role play 306 nobel physic prize 279 national medical library 273 carbon water dioxide 261 second thermodynamics law 247 speed sound of_pobj density population compound take place dobj link external compound 0.3 :(
  • 18. word2vec combinations Wanted to capture the intersection of meanings, but didn’t know how to combine word2vec representations TFIDFqa pairs Combinations of question tokens Combinations of answer tokens Cosine similarity Max score ~ 0.3 :( even with careful kw filtering word2gauss didn’t help too
  • 19. Averaging Neural networks (1st encounter) w2v_dim = 300 vec_q = mean(w2v(Q)) vec_c = mean(w2v(Ac)) vec_w = mean(w2v(Aw)) cos_sim(vec_q, vec_c) > cos_sim(vec_q, vec_w)
  • 20. Averaging Neural networks (1st encounter) w2v_dim = 300 vec_q = mean(w2v(Q)) vec_c = mean(w2v(Ac)) vec_w = mean(w2v(Aw)) a = CNN(w2v(X)) vec_x = mean(w2v(X) * a) cos_sim(vec_q, vec_c) > cos_sim(vec_q, vec_w)
  • 21. Semantic Neural networks (2nd encounter) + Paragraphs LSTM = LSTM(w2v) LSTM(s1 | s2) > LSTM(s1 | s3) if s1 and s2 are from the same paragraph, while s1 and s3 are not LSTM(a, b) is low then a and b are from the same paragraph (energy based learning) Loss = max(0, M - LSTM(s1, s2) + LSTM(s1, s3)) Score: 0.26
  • 24. Reading Neural networks (3rd encounter) + Lots of paragraphs + Search Engine + A survey: - bigrams are not accounted - main idea (keywords) of a sentence is not recognized
  • 25. Reading Neural networks (3rd encounter) + Lots of paragraphs + Search Engine + A survey: - bigrams are not accounted - main idea (keywords) of a sentence is not recognized
  • 26. Reading Neural networks (3rd encounter) All we want is to know if a sentence is from a paragraph to be able to rerank lucene scores.
  • 27.
  • 29. Reading Neural networks (3rd encounter) sentences -> LSTM -> Dense NN -> Embedding w2v -> LSTM -> Dense NN -> Embedding w2v -> Mean -> Dense NN -> Embedding
  • 30. Neural networks. Learned lessons. Start as small as possible Corruption is important for siamese networks Learning curve is misleading in NLP
  • 31. Lessons learned Start early - wasted two first months of the competition (but had a week of 24/7 hackathon at the end) No stickers in the team channel (except with Yann LeCun on a good submit) Common toolbox is nice A dedicated server is a good thing to have (no need in AWS spot instances) Experiment fast, fail early Team work means a lot