SlideShare una empresa de Scribd logo
1 de 35
Descargar para leer sin conexión
Learning to Grow Structured Visual Summaries
for Document Collections
Daniil Mirylenka Andrea Passerini
University of Trento, Italy
Machine learning seminar, Waikato University, 2013
Problem: informative representation of documents
Application: academic search
Input: document collection Output: topic map
⇒
Our approach:
Building and summarizing the topic graph
⇒ ⇒
Building the topic graph:
Overview
1. Map documents to Wikipedia articles
2. Retrieve the parent categories
3. Link categories to each other
4. Merge similar topics
5. Break cycles in the graph
Building the topic graph:
Mapping the document to Wikipedia articles
“..we propose a method of summarizing collections
of documents with concise topic hierarchies, and
show how it can be applied to visualization and
browsing of academic search results.”
⇓
“..we propose a method summarizing collections of
documents with concise [[Topic (linguistics) |topic]]
[[Hierarchy |hierarchies]], and show how it can be
applied to [[Visualization (computer graphics)
|visualization]] and [[Web browser |browsing]] of
[[List of academic databases and search engines
|academic search]] results.”
Building the topic graph:
Retrieving the parent categories
⇓
Building the topic graph:
Linking the categories
⇓
Building the topic graph:
Merging similar topics
⇓
Building the topic graph:
Breaking the cycles
⇓
Building the topic graph:
Example of an actual topic graph built from 100 abstracts
Summarizing the topic graph
Reflection
⇒
What is a summary?
- a set of nodes (topics).
Summarizing the topic graph
Reflection
⇒
What is a summary?
- a set of nodes (topics).
What is a good summary?
- ???
Summarizing the topic graph
Reflection
⇒
What is a summary?
- a set of nodes (topics).
What is a good summary?
- ???
Let’s learn from examples!
- subjective
Summarizing the topic graph
The first attempt
Structured prediction
ˆGT = arg max
GT
F(G, GT )
Summarizing the topic graph
The first attempt
Structured prediction
ˆGT = arg max
GT
F(G, GT )
Problem: evaluation on |G|
T
subgraphs
- Example:
300-node topic graph
10-node summary
Summarizing the topic graph
The first attempt
Structured prediction
ˆGT = arg max
GT
F(G, GT )
Problem: evaluation on |G|
T
subgraphs
- Example:
300-node topic graph
10-node summary
1 398 320 233 241 701 770 possible subgraphs
(1 million graphs per second ⇒ 44 311 years)
Summarizing the topic graph
Key idea
Restriction: summaries should be nested
∅ = G0 ⊂ G1 ⊂ · · · ⊂ GT
Summarizing the topic graph
Key idea
Restriction: summaries should be nested
∅ = G0 ⊂ G1 ⊂ · · · ⊂ GT
Now we can build summaries sequentially
Gt = Gt−1 ∪ {vt}
Summarizing the topic graph
Key idea
Restriction: summaries should be nested
∅ = G0 ⊂ G1 ⊂ · · · ⊂ GT
Now we can build summaries sequentially
Gt = Gt−1 ∪ {vt}
Still a supervised learning problem
- training data: summary sequences (G, G1, G2, · · · , GT )
- or topic sequences: (G, v1, v2, · · · , vT )
Learning to grow summaries
as imitation learning
Imitation learning (racing analogy)
destination: finish
sequence of states
driver’s actions (steering, etc.)
goal: copy the behaviour
Supervised Trai
Expert Trajectories
Learned Policy: aˆsup 
(borrowed from the presentation of Stephane Ross)
Learning to grow summaries
as imitation learning
Imitation learning (racing analogy)
destination: finish
sequence of states
driver’s actions (steering, etc.)
goal: copy the behaviour
Supervised Trai
Expert Trajectories
Learned Policy: aˆsup 
(borrowed from the presentation of Stephane Ross)
Our problem
destination: summary GT
states: intermediate summaries G0, G1, · · · , GT−1
actions: topics v1, v2, · · · , vT added to the summaries
goal: copy the behaviour
Learning to grow summaries
How can we do that?
Straightforward approach
Choose a classifier π : (G, Gt−1) → vt
Train on the ‘ground truth’ examples ((G, Gt−1), vt)
Sequentially apply on the new graphs
∅ = ˆG0
π(G,.)
→ ˆG1
π(G,.)
→ · · ·
π(G,.)
→ ˆGT
Learning to grow summaries
How can we do that?
Straightforward approach
Choose a classifier π : (G, Gt−1) → vt
Train on the ‘ground truth’ examples ((G, Gt−1), vt)
Sequentially apply on the new graphs
∅ = ˆG0
π(G,.)
→ ˆG1
π(G,.)
→ · · ·
π(G,.)
→ ˆGT
Will it work?
Learning to grow summaries
How can we do that?
Straightforward approach
Choose a classifier π : (G, Gt−1) → vt
Train on the ‘ground truth’ examples ((G, Gt−1), vt)
Sequentially apply on the new graphs
∅ = ˆG0
π(G,.)
→ ˆG1
π(G,.)
→ · · ·
π(G,.)
→ ˆGT
Will it work?
No.
(unable to recover from mistakes)
Learning to grow summaries
DAgger (dataset aggregation)
S. Ross, G. J. Gordon, and D. Bagnell. A reduction of imitation learning and structured prediction to no-regret
online learning. Journal of Machine Learning Research - Proceedings Track, 15:627635, 2011.
Idea:
train on the states we are going to encounter
(our own-generated states)
Learning to grow summaries
DAgger (dataset aggregation)
S. Ross, G. J. Gordon, and D. Bagnell. A reduction of imitation learning and structured prediction to no-regret
online learning. Journal of Machine Learning Research - Proceedings Track, 15:627635, 2011.
Idea:
train on the states we are going to encounter
(our own-generated states)
How can we do that?
We haven’t trained the classifier yet!
Learning to grow summaries
DAgger (dataset aggregation)
S. Ross, G. J. Gordon, and D. Bagnell. A reduction of imitation learning and structured prediction to no-regret
online learning. Journal of Machine Learning Research - Proceedings Track, 15:627635, 2011.
Idea:
train on the states we are going to encounter
(our own-generated states)
How can we do that?
We haven’t trained the classifier yet!
We will do it iteratively (for i = 0, 1,)
train the classifier πi on the dataset Di
generate the trajectories using πi
add new states to the dataset Di+1
Learning to grow summaries
Collecting the actions
DAgger (dataset aggregation)
iterating, we collect states
but we also need actions
Learning to grow summaries
Collecting the actions
DAgger (dataset aggregation)
iterating, we collect states
but we also need actions
“Let the expert steer”
Q: What action is optimal?
A: One that brings us closest to
the optimal trajectory.
DAgger: Dataset Aggregation
• Collect new trajectories with 1
1
14
Steering from
expert
(borrowed from the presentation of Stephane Ross)
Learning to grow summaries
Recap of the algorithm
The algorithm
‘ground truth’ dataset: points
(state, action)
train π on the ‘ground truth’
dataset
apply π to the initial states
- generate the trajectories
generate expert’s actions
add new state-action pairs to
the dataset
repeat
DAgger: Dataset Aggregation
• Collect new trajectories with 1
1
14
Steering from
expert
(borrowed from the presentation of Stephane Ross)
Learning to grow summaries
Training the classifier
Classifier
π : (G, Gt−1) → vt
Scoring function
F(G, Gt−1, vt) = w, Ψ (G, Gt−1, vt)
Prediction
vt = arg maxv F(G, Gt−1, v)
Learning: SVMstruct
- ensures that optimal topics score best
Learning to grow summaries
Providing the expert’s actions
Expert’s action
brings us closest to the optimal trajectory
Technically
by minimizing the loss function
vt = arg min
v
G (Gt−1 ∪ {v}, Gopt
t )
Loss functions
graphs as topic sets ⇒ redundancy
key: consider similarity between the topics
Learning grow summaries
Graph features
Some of the features:
document coverage
transitive document coverage
average and max. overlap between topics
average and max. parent-child overlap
the height of the graph
the number of connected components
...
Initial experiments
Evaluation
Microsoft Academic Search
10 manually annotated queries
leave-one-out cross-validation
greedy coverage baseline
spectral clustering-based method
based on U. Scaiella, P. Ferragina, A. Marino, and M.
Ciaramita. Topical clustering of search results. WSDM 2012.
Notes
small number of points
unique task ⇒
no established datasets
no appropriate competitor
approaches
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q q
q
q
q
q
q
q
q
q
q q
q
q q
q
q
q
q q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
q
1 2 3 4 5 6 7 8
0.20.30.40.50.60.70.8
Number n of predicted topics
match@n
q
q
q
q
q
GreedyCov
LSG
our method: 1st iteration
our method, iterations 2−9
our method, 10th iteration
Thank You
Thank You!
Questions?
Daniil Mirylenka
dmirylenka@disi.unitn.it

Más contenido relacionado

La actualidad más candente

Alternative Infill Strategies for Expensive Multi-Objective Optimisation
Alternative Infill Strategies for Expensive Multi-Objective OptimisationAlternative Infill Strategies for Expensive Multi-Objective Optimisation
Alternative Infill Strategies for Expensive Multi-Objective OptimisationAlma Rahat
 
Introduction to second gradient theory of elasticity - Arjun Narayanan
Introduction to second gradient theory of elasticity - Arjun NarayananIntroduction to second gradient theory of elasticity - Arjun Narayanan
Introduction to second gradient theory of elasticity - Arjun NarayananArjun Narayanan
 
Bayesian selection of best subsets in high-dimensional regression
Bayesian selection of best subsets in high-dimensional regressionBayesian selection of best subsets in high-dimensional regression
Bayesian selection of best subsets in high-dimensional regressionCaleb (Shiqiang) Jin
 
Supervised Prediction of Graph Summaries
Supervised Prediction of Graph SummariesSupervised Prediction of Graph Summaries
Supervised Prediction of Graph SummariesDaniil Mirylenka
 
Modeling the Dynamics of SGD by Stochastic Differential Equation
Modeling the Dynamics of SGD by Stochastic Differential EquationModeling the Dynamics of SGD by Stochastic Differential Equation
Modeling the Dynamics of SGD by Stochastic Differential EquationMark Chang
 
The Lazy Traveling Salesman Memory Management for Large-Scale Link Discovery
The Lazy Traveling Salesman Memory Management for Large-Scale Link DiscoveryThe Lazy Traveling Salesman Memory Management for Large-Scale Link Discovery
The Lazy Traveling Salesman Memory Management for Large-Scale Link DiscoveryHolistic Benchmarking of Big Linked Data
 
Nearly optimal average case complexity of counting bicliques under seth
Nearly optimal average case complexity of counting bicliques under sethNearly optimal average case complexity of counting bicliques under seth
Nearly optimal average case complexity of counting bicliques under sethNobutaka Shimizu
 
NICE Implementations of Variational Inference
NICE Implementations of Variational Inference NICE Implementations of Variational Inference
NICE Implementations of Variational Inference Natan Katz
 
Deep generative learning_icml_part1
Deep generative learning_icml_part1Deep generative learning_icml_part1
Deep generative learning_icml_part1Scyfer
 
Parallel Optimization in Machine Learning
Parallel Optimization in Machine LearningParallel Optimization in Machine Learning
Parallel Optimization in Machine LearningFabian Pedregosa
 
Quadratics in vertex form
Quadratics in vertex formQuadratics in vertex form
Quadratics in vertex formDouglas Agyei
 
From Changes to Dynamics: Dynamics Analysis of Linked Open Data Sources
From Changes to Dynamics: Dynamics Analysis of Linked Open Data Sources From Changes to Dynamics: Dynamics Analysis of Linked Open Data Sources
From Changes to Dynamics: Dynamics Analysis of Linked Open Data Sources Thomas Gottron
 
An overview of Bayesian testing
An overview of Bayesian testingAn overview of Bayesian testing
An overview of Bayesian testingChristian Robert
 
Paper Review: An exact mapping between the Variational Renormalization Group ...
Paper Review: An exact mapping between the Variational Renormalization Group ...Paper Review: An exact mapping between the Variational Renormalization Group ...
Paper Review: An exact mapping between the Variational Renormalization Group ...Kai-Wen Zhao
 
A bayesian interpretation of interpolated kneser ney
A bayesian interpretation of interpolated kneser ney A bayesian interpretation of interpolated kneser ney
A bayesian interpretation of interpolated kneser ney Satya Vasanth Reddy T
 

La actualidad más candente (19)

Alternative Infill Strategies for Expensive Multi-Objective Optimisation
Alternative Infill Strategies for Expensive Multi-Objective OptimisationAlternative Infill Strategies for Expensive Multi-Objective Optimisation
Alternative Infill Strategies for Expensive Multi-Objective Optimisation
 
Introduction to second gradient theory of elasticity - Arjun Narayanan
Introduction to second gradient theory of elasticity - Arjun NarayananIntroduction to second gradient theory of elasticity - Arjun Narayanan
Introduction to second gradient theory of elasticity - Arjun Narayanan
 
Bayesian selection of best subsets in high-dimensional regression
Bayesian selection of best subsets in high-dimensional regressionBayesian selection of best subsets in high-dimensional regression
Bayesian selection of best subsets in high-dimensional regression
 
Supervised Prediction of Graph Summaries
Supervised Prediction of Graph SummariesSupervised Prediction of Graph Summaries
Supervised Prediction of Graph Summaries
 
report
reportreport
report
 
Modeling the Dynamics of SGD by Stochastic Differential Equation
Modeling the Dynamics of SGD by Stochastic Differential EquationModeling the Dynamics of SGD by Stochastic Differential Equation
Modeling the Dynamics of SGD by Stochastic Differential Equation
 
The Lazy Traveling Salesman Memory Management for Large-Scale Link Discovery
The Lazy Traveling Salesman Memory Management for Large-Scale Link DiscoveryThe Lazy Traveling Salesman Memory Management for Large-Scale Link Discovery
The Lazy Traveling Salesman Memory Management for Large-Scale Link Discovery
 
Nearly optimal average case complexity of counting bicliques under seth
Nearly optimal average case complexity of counting bicliques under sethNearly optimal average case complexity of counting bicliques under seth
Nearly optimal average case complexity of counting bicliques under seth
 
NICE Implementations of Variational Inference
NICE Implementations of Variational Inference NICE Implementations of Variational Inference
NICE Implementations of Variational Inference
 
Deep generative learning_icml_part1
Deep generative learning_icml_part1Deep generative learning_icml_part1
Deep generative learning_icml_part1
 
Deep Learning Opening Workshop - Domain Adaptation Challenges in Genomics: a ...
Deep Learning Opening Workshop - Domain Adaptation Challenges in Genomics: a ...Deep Learning Opening Workshop - Domain Adaptation Challenges in Genomics: a ...
Deep Learning Opening Workshop - Domain Adaptation Challenges in Genomics: a ...
 
Parallel Optimization in Machine Learning
Parallel Optimization in Machine LearningParallel Optimization in Machine Learning
Parallel Optimization in Machine Learning
 
Deep Learning Opening Workshop - Deep ReLU Networks Viewed as a Statistical M...
Deep Learning Opening Workshop - Deep ReLU Networks Viewed as a Statistical M...Deep Learning Opening Workshop - Deep ReLU Networks Viewed as a Statistical M...
Deep Learning Opening Workshop - Deep ReLU Networks Viewed as a Statistical M...
 
Quadratics in vertex form
Quadratics in vertex formQuadratics in vertex form
Quadratics in vertex form
 
From Changes to Dynamics: Dynamics Analysis of Linked Open Data Sources
From Changes to Dynamics: Dynamics Analysis of Linked Open Data Sources From Changes to Dynamics: Dynamics Analysis of Linked Open Data Sources
From Changes to Dynamics: Dynamics Analysis of Linked Open Data Sources
 
An overview of Bayesian testing
An overview of Bayesian testingAn overview of Bayesian testing
An overview of Bayesian testing
 
Paper Review: An exact mapping between the Variational Renormalization Group ...
Paper Review: An exact mapping between the Variational Renormalization Group ...Paper Review: An exact mapping between the Variational Renormalization Group ...
Paper Review: An exact mapping between the Variational Renormalization Group ...
 
A bayesian interpretation of interpolated kneser ney
A bayesian interpretation of interpolated kneser ney A bayesian interpretation of interpolated kneser ney
A bayesian interpretation of interpolated kneser ney
 
Recursive algorithms
Recursive algorithmsRecursive algorithms
Recursive algorithms
 

Destacado

Леммл т. Ccna. cisco certified network associate. учебное руководство (2 е из...
Леммл т. Ccna. cisco certified network associate. учебное руководство (2 е из...Леммл т. Ccna. cisco certified network associate. учебное руководство (2 е из...
Леммл т. Ccna. cisco certified network associate. учебное руководство (2 е из...StAlKeRoV
 
Kesadaran Diri dan Afek Emosi Manusia
Kesadaran Diri dan Afek Emosi ManusiaKesadaran Diri dan Afek Emosi Manusia
Kesadaran Diri dan Afek Emosi Manusiapjj_kemenkes
 
Truly Proud!-Article on Mentorship 2015
Truly Proud!-Article on Mentorship 2015Truly Proud!-Article on Mentorship 2015
Truly Proud!-Article on Mentorship 2015Jason Love
 
Blog of the senior hindi journalists a comparitive analysis
Blog of the senior hindi journalists   a comparitive analysisBlog of the senior hindi journalists   a comparitive analysis
Blog of the senior hindi journalists a comparitive analysisAshish Kumar
 
resume final
resume finalresume final
resume finalSam Grove
 
Du lịch singapore malaysia indonesia hoàn toàn mới - khởi hành hà nội-tp.hcm
Du lịch singapore malaysia indonesia   hoàn toàn mới - khởi hành hà nội-tp.hcmDu lịch singapore malaysia indonesia   hoàn toàn mới - khởi hành hà nội-tp.hcm
Du lịch singapore malaysia indonesia hoàn toàn mới - khởi hành hà nội-tp.hcmCông ty du lịch lữ hành Saoviettravel
 
Aprendizaje autonomo
Aprendizaje autonomoAprendizaje autonomo
Aprendizaje autonomogabo2409
 
Crisis Communications Planning in the Social Media Age - On-Demand Webinar
Crisis Communications Planning in the Social Media Age - On-Demand WebinarCrisis Communications Planning in the Social Media Age - On-Demand Webinar
Crisis Communications Planning in the Social Media Age - On-Demand WebinarCNW Group
 

Destacado (13)

Леммл т. Ccna. cisco certified network associate. учебное руководство (2 е из...
Леммл т. Ccna. cisco certified network associate. учебное руководство (2 е из...Леммл т. Ccna. cisco certified network associate. учебное руководство (2 е из...
Леммл т. Ccna. cisco certified network associate. учебное руководство (2 е из...
 
Gia su gia dinh
Gia su gia dinhGia su gia dinh
Gia su gia dinh
 
Vier mythes over geherfabriceerde cartridges ontkracht
Vier mythes over geherfabriceerde cartridges ontkrachtVier mythes over geherfabriceerde cartridges ontkracht
Vier mythes over geherfabriceerde cartridges ontkracht
 
Kesadaran Diri dan Afek Emosi Manusia
Kesadaran Diri dan Afek Emosi ManusiaKesadaran Diri dan Afek Emosi Manusia
Kesadaran Diri dan Afek Emosi Manusia
 
Truly Proud!-Article on Mentorship 2015
Truly Proud!-Article on Mentorship 2015Truly Proud!-Article on Mentorship 2015
Truly Proud!-Article on Mentorship 2015
 
αλγόριθμοι 2.2
αλγόριθμοι 2.2αλγόριθμοι 2.2
αλγόριθμοι 2.2
 
Blog of the senior hindi journalists a comparitive analysis
Blog of the senior hindi journalists   a comparitive analysisBlog of the senior hindi journalists   a comparitive analysis
Blog of the senior hindi journalists a comparitive analysis
 
Monroe
MonroeMonroe
Monroe
 
resume final
resume finalresume final
resume final
 
Du lịch singapore malaysia indonesia hoàn toàn mới - khởi hành hà nội-tp.hcm
Du lịch singapore malaysia indonesia   hoàn toàn mới - khởi hành hà nội-tp.hcmDu lịch singapore malaysia indonesia   hoàn toàn mới - khởi hành hà nội-tp.hcm
Du lịch singapore malaysia indonesia hoàn toàn mới - khởi hành hà nội-tp.hcm
 
Aprendizaje autonomo
Aprendizaje autonomoAprendizaje autonomo
Aprendizaje autonomo
 
Crisis Communications Planning in the Social Media Age - On-Demand Webinar
Crisis Communications Planning in the Social Media Age - On-Demand WebinarCrisis Communications Planning in the Social Media Age - On-Demand Webinar
Crisis Communications Planning in the Social Media Age - On-Demand Webinar
 
LF1643774 (Unilever)
LF1643774 (Unilever)LF1643774 (Unilever)
LF1643774 (Unilever)
 

Similar a Learning to Grow Structured Visual Summaries for Document Collections

AI – Week 21 Machine Learning: Macro Learning
AI – Week 21 Machine Learning: Macro LearningAI – Week 21 Machine Learning: Macro Learning
AI – Week 21 Machine Learning: Macro Learningbutest
 
Chap 8. Optimization for training deep models
Chap 8. Optimization for training deep modelsChap 8. Optimization for training deep models
Chap 8. Optimization for training deep modelsYoung-Geun Choi
 
08 distributed optimization
08 distributed optimization08 distributed optimization
08 distributed optimizationMarco Quartulli
 
Online advertising and large scale model fitting
Online advertising and large scale model fittingOnline advertising and large scale model fitting
Online advertising and large scale model fittingWush Wu
 
Rohan's Masters presentation
Rohan's Masters presentationRohan's Masters presentation
Rohan's Masters presentationrohan_anil
 
Machine learning ppt unit one syllabuspptx
Machine learning ppt unit one syllabuspptxMachine learning ppt unit one syllabuspptx
Machine learning ppt unit one syllabuspptxVenkateswaraBabuRavi
 
Reinforcement Learning (Reloaded) - Xavier Giró-i-Nieto - UPC Barcelona 2018
Reinforcement Learning (Reloaded) - Xavier Giró-i-Nieto - UPC Barcelona 2018Reinforcement Learning (Reloaded) - Xavier Giró-i-Nieto - UPC Barcelona 2018
Reinforcement Learning (Reloaded) - Xavier Giró-i-Nieto - UPC Barcelona 2018Universitat Politècnica de Catalunya
 
Generative modeling with Convolutional Neural Networks
Generative modeling with Convolutional Neural NetworksGenerative modeling with Convolutional Neural Networks
Generative modeling with Convolutional Neural NetworksDenis Dus
 
Multi-scalar multiplication: state of the art and new ideas
Multi-scalar multiplication: state of the art and new ideasMulti-scalar multiplication: state of the art and new ideas
Multi-scalar multiplication: state of the art and new ideasGus Gutoski
 
Learning to Rank with Neural Networks
Learning to Rank with Neural NetworksLearning to Rank with Neural Networks
Learning to Rank with Neural NetworksBhaskar Mitra
 
Ch 9-1.Machine Learning: Symbol-based
Ch 9-1.Machine Learning: Symbol-basedCh 9-1.Machine Learning: Symbol-based
Ch 9-1.Machine Learning: Symbol-basedbutest
 
Ch 9-1.Machine Learning: Symbol-based
Ch 9-1.Machine Learning: Symbol-basedCh 9-1.Machine Learning: Symbol-based
Ch 9-1.Machine Learning: Symbol-basedbutest
 
Joint contrastive learning with infinite possibilities
Joint contrastive learning with infinite possibilitiesJoint contrastive learning with infinite possibilities
Joint contrastive learning with infinite possibilitiestaeseon ryu
 
Classes without Dependencies - UseR 2018
Classes without Dependencies - UseR 2018Classes without Dependencies - UseR 2018
Classes without Dependencies - UseR 2018Sam Clifford
 
learning boolean weight learning real valued weights rank learning as ordina...
learning boolean weight learning real valued weights  rank learning as ordina...learning boolean weight learning real valued weights  rank learning as ordina...
learning boolean weight learning real valued weights rank learning as ordina...jaishriramm0
 
How to apply deep learning to 3 d objects
How to apply deep learning to 3 d objectsHow to apply deep learning to 3 d objects
How to apply deep learning to 3 d objectsOgushi Masaya
 

Similar a Learning to Grow Structured Visual Summaries for Document Collections (20)

AI – Week 21 Machine Learning: Macro Learning
AI – Week 21 Machine Learning: Macro LearningAI – Week 21 Machine Learning: Macro Learning
AI – Week 21 Machine Learning: Macro Learning
 
Exhaustive Combinatorial Enumeration
Exhaustive Combinatorial EnumerationExhaustive Combinatorial Enumeration
Exhaustive Combinatorial Enumeration
 
04 numerical
04 numerical04 numerical
04 numerical
 
Chap 8. Optimization for training deep models
Chap 8. Optimization for training deep modelsChap 8. Optimization for training deep models
Chap 8. Optimization for training deep models
 
08 distributed optimization
08 distributed optimization08 distributed optimization
08 distributed optimization
 
Online advertising and large scale model fitting
Online advertising and large scale model fittingOnline advertising and large scale model fitting
Online advertising and large scale model fitting
 
Rohan's Masters presentation
Rohan's Masters presentationRohan's Masters presentation
Rohan's Masters presentation
 
Machine learning ppt unit one syllabuspptx
Machine learning ppt unit one syllabuspptxMachine learning ppt unit one syllabuspptx
Machine learning ppt unit one syllabuspptx
 
Reinforcement Learning (Reloaded) - Xavier Giró-i-Nieto - UPC Barcelona 2018
Reinforcement Learning (Reloaded) - Xavier Giró-i-Nieto - UPC Barcelona 2018Reinforcement Learning (Reloaded) - Xavier Giró-i-Nieto - UPC Barcelona 2018
Reinforcement Learning (Reloaded) - Xavier Giró-i-Nieto - UPC Barcelona 2018
 
Learning To Run
Learning To RunLearning To Run
Learning To Run
 
Generative modeling with Convolutional Neural Networks
Generative modeling with Convolutional Neural NetworksGenerative modeling with Convolutional Neural Networks
Generative modeling with Convolutional Neural Networks
 
Multi-scalar multiplication: state of the art and new ideas
Multi-scalar multiplication: state of the art and new ideasMulti-scalar multiplication: state of the art and new ideas
Multi-scalar multiplication: state of the art and new ideas
 
Learning to Rank with Neural Networks
Learning to Rank with Neural NetworksLearning to Rank with Neural Networks
Learning to Rank with Neural Networks
 
Ch 9-1.Machine Learning: Symbol-based
Ch 9-1.Machine Learning: Symbol-basedCh 9-1.Machine Learning: Symbol-based
Ch 9-1.Machine Learning: Symbol-based
 
Ch 9-1.Machine Learning: Symbol-based
Ch 9-1.Machine Learning: Symbol-basedCh 9-1.Machine Learning: Symbol-based
Ch 9-1.Machine Learning: Symbol-based
 
Joint contrastive learning with infinite possibilities
Joint contrastive learning with infinite possibilitiesJoint contrastive learning with infinite possibilities
Joint contrastive learning with infinite possibilities
 
Classes without Dependencies - UseR 2018
Classes without Dependencies - UseR 2018Classes without Dependencies - UseR 2018
Classes without Dependencies - UseR 2018
 
Interval Pattern Structures: An introdution
Interval Pattern Structures: An introdutionInterval Pattern Structures: An introdution
Interval Pattern Structures: An introdution
 
learning boolean weight learning real valued weights rank learning as ordina...
learning boolean weight learning real valued weights  rank learning as ordina...learning boolean weight learning real valued weights  rank learning as ordina...
learning boolean weight learning real valued weights rank learning as ordina...
 
How to apply deep learning to 3 d objects
How to apply deep learning to 3 d objectsHow to apply deep learning to 3 d objects
How to apply deep learning to 3 d objects
 

Último

Landscape Catalogue 2024 Australia-1.pdf
Landscape Catalogue 2024 Australia-1.pdfLandscape Catalogue 2024 Australia-1.pdf
Landscape Catalogue 2024 Australia-1.pdfAarwolf Industries LLC
 
The Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptx
The Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptxThe Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptx
The Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptxLoriGlavin3
 
Microservices, Docker deploy and Microservices source code in C#
Microservices, Docker deploy and Microservices source code in C#Microservices, Docker deploy and Microservices source code in C#
Microservices, Docker deploy and Microservices source code in C#Karmanjay Verma
 
Generative Artificial Intelligence: How generative AI works.pdf
Generative Artificial Intelligence: How generative AI works.pdfGenerative Artificial Intelligence: How generative AI works.pdf
Generative Artificial Intelligence: How generative AI works.pdfIngrid Airi González
 
A Deep Dive on Passkeys: FIDO Paris Seminar.pptx
A Deep Dive on Passkeys: FIDO Paris Seminar.pptxA Deep Dive on Passkeys: FIDO Paris Seminar.pptx
A Deep Dive on Passkeys: FIDO Paris Seminar.pptxLoriGlavin3
 
QCon London: Mastering long-running processes in modern architectures
QCon London: Mastering long-running processes in modern architecturesQCon London: Mastering long-running processes in modern architectures
QCon London: Mastering long-running processes in modern architecturesBernd Ruecker
 
A Framework for Development in the AI Age
A Framework for Development in the AI AgeA Framework for Development in the AI Age
A Framework for Development in the AI AgeCprime
 
All These Sophisticated Attacks, Can We Really Detect Them - PDF
All These Sophisticated Attacks, Can We Really Detect Them - PDFAll These Sophisticated Attacks, Can We Really Detect Them - PDF
All These Sophisticated Attacks, Can We Really Detect Them - PDFMichael Gough
 
Testing tools and AI - ideas what to try with some tool examples
Testing tools and AI - ideas what to try with some tool examplesTesting tools and AI - ideas what to try with some tool examples
Testing tools and AI - ideas what to try with some tool examplesKari Kakkonen
 
Varsha Sewlal- Cyber Attacks on Critical Critical Infrastructure
Varsha Sewlal- Cyber Attacks on Critical Critical InfrastructureVarsha Sewlal- Cyber Attacks on Critical Critical Infrastructure
Varsha Sewlal- Cyber Attacks on Critical Critical Infrastructureitnewsafrica
 
Emixa Mendix Meetup 11 April 2024 about Mendix Native development
Emixa Mendix Meetup 11 April 2024 about Mendix Native developmentEmixa Mendix Meetup 11 April 2024 about Mendix Native development
Emixa Mendix Meetup 11 April 2024 about Mendix Native developmentPim van der Noll
 
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...Alkin Tezuysal
 
So einfach geht modernes Roaming fuer Notes und Nomad.pdf
So einfach geht modernes Roaming fuer Notes und Nomad.pdfSo einfach geht modernes Roaming fuer Notes und Nomad.pdf
So einfach geht modernes Roaming fuer Notes und Nomad.pdfpanagenda
 
Glenn Lazarus- Why Your Observability Strategy Needs Security Observability
Glenn Lazarus- Why Your Observability Strategy Needs Security ObservabilityGlenn Lazarus- Why Your Observability Strategy Needs Security Observability
Glenn Lazarus- Why Your Observability Strategy Needs Security Observabilityitnewsafrica
 
A Glance At The Java Performance Toolbox
A Glance At The Java Performance ToolboxA Glance At The Java Performance Toolbox
A Glance At The Java Performance ToolboxAna-Maria Mihalceanu
 
React JS; all concepts. Contains React Features, JSX, functional & Class comp...
React JS; all concepts. Contains React Features, JSX, functional & Class comp...React JS; all concepts. Contains React Features, JSX, functional & Class comp...
React JS; all concepts. Contains React Features, JSX, functional & Class comp...Karmanjay Verma
 
2024 April Patch Tuesday
2024 April Patch Tuesday2024 April Patch Tuesday
2024 April Patch TuesdayIvanti
 
Data governance with Unity Catalog Presentation
Data governance with Unity Catalog PresentationData governance with Unity Catalog Presentation
Data governance with Unity Catalog PresentationKnoldus Inc.
 
[Webinar] SpiraTest - Setting New Standards in Quality Assurance
[Webinar] SpiraTest - Setting New Standards in Quality Assurance[Webinar] SpiraTest - Setting New Standards in Quality Assurance
[Webinar] SpiraTest - Setting New Standards in Quality AssuranceInflectra
 
React Native vs Ionic - The Best Mobile App Framework
React Native vs Ionic - The Best Mobile App FrameworkReact Native vs Ionic - The Best Mobile App Framework
React Native vs Ionic - The Best Mobile App FrameworkPixlogix Infotech
 

Último (20)

Landscape Catalogue 2024 Australia-1.pdf
Landscape Catalogue 2024 Australia-1.pdfLandscape Catalogue 2024 Australia-1.pdf
Landscape Catalogue 2024 Australia-1.pdf
 
The Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptx
The Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptxThe Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptx
The Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptx
 
Microservices, Docker deploy and Microservices source code in C#
Microservices, Docker deploy and Microservices source code in C#Microservices, Docker deploy and Microservices source code in C#
Microservices, Docker deploy and Microservices source code in C#
 
Generative Artificial Intelligence: How generative AI works.pdf
Generative Artificial Intelligence: How generative AI works.pdfGenerative Artificial Intelligence: How generative AI works.pdf
Generative Artificial Intelligence: How generative AI works.pdf
 
A Deep Dive on Passkeys: FIDO Paris Seminar.pptx
A Deep Dive on Passkeys: FIDO Paris Seminar.pptxA Deep Dive on Passkeys: FIDO Paris Seminar.pptx
A Deep Dive on Passkeys: FIDO Paris Seminar.pptx
 
QCon London: Mastering long-running processes in modern architectures
QCon London: Mastering long-running processes in modern architecturesQCon London: Mastering long-running processes in modern architectures
QCon London: Mastering long-running processes in modern architectures
 
A Framework for Development in the AI Age
A Framework for Development in the AI AgeA Framework for Development in the AI Age
A Framework for Development in the AI Age
 
All These Sophisticated Attacks, Can We Really Detect Them - PDF
All These Sophisticated Attacks, Can We Really Detect Them - PDFAll These Sophisticated Attacks, Can We Really Detect Them - PDF
All These Sophisticated Attacks, Can We Really Detect Them - PDF
 
Testing tools and AI - ideas what to try with some tool examples
Testing tools and AI - ideas what to try with some tool examplesTesting tools and AI - ideas what to try with some tool examples
Testing tools and AI - ideas what to try with some tool examples
 
Varsha Sewlal- Cyber Attacks on Critical Critical Infrastructure
Varsha Sewlal- Cyber Attacks on Critical Critical InfrastructureVarsha Sewlal- Cyber Attacks on Critical Critical Infrastructure
Varsha Sewlal- Cyber Attacks on Critical Critical Infrastructure
 
Emixa Mendix Meetup 11 April 2024 about Mendix Native development
Emixa Mendix Meetup 11 April 2024 about Mendix Native developmentEmixa Mendix Meetup 11 April 2024 about Mendix Native development
Emixa Mendix Meetup 11 April 2024 about Mendix Native development
 
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...
 
So einfach geht modernes Roaming fuer Notes und Nomad.pdf
So einfach geht modernes Roaming fuer Notes und Nomad.pdfSo einfach geht modernes Roaming fuer Notes und Nomad.pdf
So einfach geht modernes Roaming fuer Notes und Nomad.pdf
 
Glenn Lazarus- Why Your Observability Strategy Needs Security Observability
Glenn Lazarus- Why Your Observability Strategy Needs Security ObservabilityGlenn Lazarus- Why Your Observability Strategy Needs Security Observability
Glenn Lazarus- Why Your Observability Strategy Needs Security Observability
 
A Glance At The Java Performance Toolbox
A Glance At The Java Performance ToolboxA Glance At The Java Performance Toolbox
A Glance At The Java Performance Toolbox
 
React JS; all concepts. Contains React Features, JSX, functional & Class comp...
React JS; all concepts. Contains React Features, JSX, functional & Class comp...React JS; all concepts. Contains React Features, JSX, functional & Class comp...
React JS; all concepts. Contains React Features, JSX, functional & Class comp...
 
2024 April Patch Tuesday
2024 April Patch Tuesday2024 April Patch Tuesday
2024 April Patch Tuesday
 
Data governance with Unity Catalog Presentation
Data governance with Unity Catalog PresentationData governance with Unity Catalog Presentation
Data governance with Unity Catalog Presentation
 
[Webinar] SpiraTest - Setting New Standards in Quality Assurance
[Webinar] SpiraTest - Setting New Standards in Quality Assurance[Webinar] SpiraTest - Setting New Standards in Quality Assurance
[Webinar] SpiraTest - Setting New Standards in Quality Assurance
 
React Native vs Ionic - The Best Mobile App Framework
React Native vs Ionic - The Best Mobile App FrameworkReact Native vs Ionic - The Best Mobile App Framework
React Native vs Ionic - The Best Mobile App Framework
 

Learning to Grow Structured Visual Summaries for Document Collections

  • 1. Learning to Grow Structured Visual Summaries for Document Collections Daniil Mirylenka Andrea Passerini University of Trento, Italy Machine learning seminar, Waikato University, 2013
  • 2. Problem: informative representation of documents Application: academic search Input: document collection Output: topic map ⇒
  • 3. Our approach: Building and summarizing the topic graph ⇒ ⇒
  • 4. Building the topic graph: Overview 1. Map documents to Wikipedia articles 2. Retrieve the parent categories 3. Link categories to each other 4. Merge similar topics 5. Break cycles in the graph
  • 5. Building the topic graph: Mapping the document to Wikipedia articles “..we propose a method of summarizing collections of documents with concise topic hierarchies, and show how it can be applied to visualization and browsing of academic search results.” ⇓ “..we propose a method summarizing collections of documents with concise [[Topic (linguistics) |topic]] [[Hierarchy |hierarchies]], and show how it can be applied to [[Visualization (computer graphics) |visualization]] and [[Web browser |browsing]] of [[List of academic databases and search engines |academic search]] results.”
  • 6. Building the topic graph: Retrieving the parent categories ⇓
  • 7. Building the topic graph: Linking the categories ⇓
  • 8. Building the topic graph: Merging similar topics ⇓
  • 9. Building the topic graph: Breaking the cycles ⇓
  • 10. Building the topic graph: Example of an actual topic graph built from 100 abstracts
  • 11. Summarizing the topic graph Reflection ⇒ What is a summary? - a set of nodes (topics).
  • 12. Summarizing the topic graph Reflection ⇒ What is a summary? - a set of nodes (topics). What is a good summary? - ???
  • 13. Summarizing the topic graph Reflection ⇒ What is a summary? - a set of nodes (topics). What is a good summary? - ??? Let’s learn from examples! - subjective
  • 14. Summarizing the topic graph The first attempt Structured prediction ˆGT = arg max GT F(G, GT )
  • 15. Summarizing the topic graph The first attempt Structured prediction ˆGT = arg max GT F(G, GT ) Problem: evaluation on |G| T subgraphs - Example: 300-node topic graph 10-node summary
  • 16. Summarizing the topic graph The first attempt Structured prediction ˆGT = arg max GT F(G, GT ) Problem: evaluation on |G| T subgraphs - Example: 300-node topic graph 10-node summary 1 398 320 233 241 701 770 possible subgraphs (1 million graphs per second ⇒ 44 311 years)
  • 17. Summarizing the topic graph Key idea Restriction: summaries should be nested ∅ = G0 ⊂ G1 ⊂ · · · ⊂ GT
  • 18. Summarizing the topic graph Key idea Restriction: summaries should be nested ∅ = G0 ⊂ G1 ⊂ · · · ⊂ GT Now we can build summaries sequentially Gt = Gt−1 ∪ {vt}
  • 19. Summarizing the topic graph Key idea Restriction: summaries should be nested ∅ = G0 ⊂ G1 ⊂ · · · ⊂ GT Now we can build summaries sequentially Gt = Gt−1 ∪ {vt} Still a supervised learning problem - training data: summary sequences (G, G1, G2, · · · , GT ) - or topic sequences: (G, v1, v2, · · · , vT )
  • 20. Learning to grow summaries as imitation learning Imitation learning (racing analogy) destination: finish sequence of states driver’s actions (steering, etc.) goal: copy the behaviour Supervised Trai Expert Trajectories Learned Policy: aˆsup  (borrowed from the presentation of Stephane Ross)
  • 21. Learning to grow summaries as imitation learning Imitation learning (racing analogy) destination: finish sequence of states driver’s actions (steering, etc.) goal: copy the behaviour Supervised Trai Expert Trajectories Learned Policy: aˆsup  (borrowed from the presentation of Stephane Ross) Our problem destination: summary GT states: intermediate summaries G0, G1, · · · , GT−1 actions: topics v1, v2, · · · , vT added to the summaries goal: copy the behaviour
  • 22. Learning to grow summaries How can we do that? Straightforward approach Choose a classifier π : (G, Gt−1) → vt Train on the ‘ground truth’ examples ((G, Gt−1), vt) Sequentially apply on the new graphs ∅ = ˆG0 π(G,.) → ˆG1 π(G,.) → · · · π(G,.) → ˆGT
  • 23. Learning to grow summaries How can we do that? Straightforward approach Choose a classifier π : (G, Gt−1) → vt Train on the ‘ground truth’ examples ((G, Gt−1), vt) Sequentially apply on the new graphs ∅ = ˆG0 π(G,.) → ˆG1 π(G,.) → · · · π(G,.) → ˆGT Will it work?
  • 24. Learning to grow summaries How can we do that? Straightforward approach Choose a classifier π : (G, Gt−1) → vt Train on the ‘ground truth’ examples ((G, Gt−1), vt) Sequentially apply on the new graphs ∅ = ˆG0 π(G,.) → ˆG1 π(G,.) → · · · π(G,.) → ˆGT Will it work? No. (unable to recover from mistakes)
  • 25. Learning to grow summaries DAgger (dataset aggregation) S. Ross, G. J. Gordon, and D. Bagnell. A reduction of imitation learning and structured prediction to no-regret online learning. Journal of Machine Learning Research - Proceedings Track, 15:627635, 2011. Idea: train on the states we are going to encounter (our own-generated states)
  • 26. Learning to grow summaries DAgger (dataset aggregation) S. Ross, G. J. Gordon, and D. Bagnell. A reduction of imitation learning and structured prediction to no-regret online learning. Journal of Machine Learning Research - Proceedings Track, 15:627635, 2011. Idea: train on the states we are going to encounter (our own-generated states) How can we do that? We haven’t trained the classifier yet!
  • 27. Learning to grow summaries DAgger (dataset aggregation) S. Ross, G. J. Gordon, and D. Bagnell. A reduction of imitation learning and structured prediction to no-regret online learning. Journal of Machine Learning Research - Proceedings Track, 15:627635, 2011. Idea: train on the states we are going to encounter (our own-generated states) How can we do that? We haven’t trained the classifier yet! We will do it iteratively (for i = 0, 1,) train the classifier πi on the dataset Di generate the trajectories using πi add new states to the dataset Di+1
  • 28. Learning to grow summaries Collecting the actions DAgger (dataset aggregation) iterating, we collect states but we also need actions
  • 29. Learning to grow summaries Collecting the actions DAgger (dataset aggregation) iterating, we collect states but we also need actions “Let the expert steer” Q: What action is optimal? A: One that brings us closest to the optimal trajectory. DAgger: Dataset Aggregation • Collect new trajectories with 1 1 14 Steering from expert (borrowed from the presentation of Stephane Ross)
  • 30. Learning to grow summaries Recap of the algorithm The algorithm ‘ground truth’ dataset: points (state, action) train π on the ‘ground truth’ dataset apply π to the initial states - generate the trajectories generate expert’s actions add new state-action pairs to the dataset repeat DAgger: Dataset Aggregation • Collect new trajectories with 1 1 14 Steering from expert (borrowed from the presentation of Stephane Ross)
  • 31. Learning to grow summaries Training the classifier Classifier π : (G, Gt−1) → vt Scoring function F(G, Gt−1, vt) = w, Ψ (G, Gt−1, vt) Prediction vt = arg maxv F(G, Gt−1, v) Learning: SVMstruct - ensures that optimal topics score best
  • 32. Learning to grow summaries Providing the expert’s actions Expert’s action brings us closest to the optimal trajectory Technically by minimizing the loss function vt = arg min v G (Gt−1 ∪ {v}, Gopt t ) Loss functions graphs as topic sets ⇒ redundancy key: consider similarity between the topics
  • 33. Learning grow summaries Graph features Some of the features: document coverage transitive document coverage average and max. overlap between topics average and max. parent-child overlap the height of the graph the number of connected components ...
  • 34. Initial experiments Evaluation Microsoft Academic Search 10 manually annotated queries leave-one-out cross-validation greedy coverage baseline spectral clustering-based method based on U. Scaiella, P. Ferragina, A. Marino, and M. Ciaramita. Topical clustering of search results. WSDM 2012. Notes small number of points unique task ⇒ no established datasets no appropriate competitor approaches q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q q 1 2 3 4 5 6 7 8 0.20.30.40.50.60.70.8 Number n of predicted topics match@n q q q q q GreedyCov LSG our method: 1st iteration our method, iterations 2−9 our method, 10th iteration
  • 35. Thank You Thank You! Questions? Daniil Mirylenka dmirylenka@disi.unitn.it