SlideShare a Scribd company logo
1 of 22
Download to read offline
Weblab@The University of Tokyo
Procedural Modeling Using Autoencoder Networks
Yumer, M. E., Asente, P., Mech, R., and Kara, L. B.
Proceedings of the ACM Symposium on User Interface Software and Technology (UIST), 2015.
Presenter: Shuhei Iitsuka
Weblab@The University of Tokyo
Introduction: Problem
Procedural modeling (PM) … Geometry modeling created by a set of parameters and algorithms.
PM provides rich representation, but it’s difficult to configure a large number of parameters.
→ More intuitive exploration method is required.
Example of PM: Tree modeling with L-system.
L-system, Wikipedia
Screenshot from a conventional PM interface.
[Figure 7 in paper]
!?!?
2
Weblab@The University of Tokyo
Introduction: Contribution
Proposed a dimensionality reduction method for PM using autoencoders.
An augmentation for continuous latent space generation (without using variational inference!) is also proposed.
Contributions:
● A method to create a lower-dimensional representations of PM using autoencoders.
● A PM design system interface using an explore-and-select interaction.
3
Screenshot from the PM design system interface.
[Figure 3 in paper]
Weblab@The University of Tokyo
Why this paper?
● Future extraction should be a key component for generating variations of web design.
● Autoencoders are attracting great attention for future extraction, so I was searching a good application of them
in the field of visual design.
● When I queried “visual design” and “autoencoder” on Google Scholar, this paper was the top hit.
4
Weblab@The University of Tokyo
Agenda
1. Related Works
2. Proposed Method
3. Evaluation and Results
4. Discussion & Conclusion
5
Weblab@The University of Tokyo
Related Works: Procedural Modeling
From when L-system is proposed, procedural modeling (PM) has many applications in digital design.
In order to migrate the burden of parameter adjustment, two approaches are proposed.
● Targeted design
● Exploratory system
6
Cities modeled by procedural modeling.
[Yoav 2001]
Weblab@The University of Tokyo
Related Works: Workaround for Parameter Adjustment
Targeted Design
Streamlines parameter adjustment by providing high-level
design grammar.
PROS: Users are free from adjusting PM parameters
directly
CONS: Less user control over the generated models.
7
Exploratory System
By providing helpful features for exploration, ease the
burden to adjust the parameters (e.g. show frequently
used models, devised parameter dialogue)
PROS: Users can start modeling from their liking initial
model.
CONS: Users need to back to the conventional PM
approach to tweak the parameters.
Design grammar for plant modeling.
[Lintermann 1999]
Visualization of goodness estimated by using crowdsourcing.
[Koyama 2014]
Weblab@The University of Tokyo
Related Works: Deep Neural Networks and Autoencoders
Deep neural networks model abstract characteristics of data using nonlinear transformations.
An autoencoder is a special form of a deep neural network which tries to reproduce the input layer at the output layer.
Autoencoders is superior to other dimensionality reduction methods [Hinton 2006].
→ Employ autoencoders for dimensionality reduction in this research.
8
Weblab@The University of Tokyo
Proposed Method: User Experience
9
Screenshot from the proposed exploration system interface.
[Figure 2 in paper]
Weblab@The University of Tokyo
Proposed Model: Sampling the Training Data
● Authors employ Categorization tree (C-tree) in order to sample data uniformly with respect to shapes rather
than parameters.
○ Create quartets of shaped based on topological structure.
○ Construct the hierarchical tree assembling the quartets.
● After the initial tree is constructed, conduct subsampling
As denoted with red rectangles in the figure.
● Iterates (1) adding new sample, (2) computing a new C-tree,
(3) performing subsampling until the criteria is satisfied.
10
Example of C-tree.
[Figure 4 in paper]
Example of topological quartet.
[Huang 2013]
Close
Far
Weblab@The University of Tokyo
Proposed Method: Neural Network Architecture
● 5-hidden-layer symmetric autoencoder.
● In addition to PM parameters, shape features are included in the input layer in order to ensure geometric continuity.
○ A silhouette-based histogram extracted from Light Field Descriptors is utilized.
● Training is done layer-by-layer followed by a fine tuning of the entire structure.
● Weighted reconstruction error on shape features is applied to the first hidden layer.
11
Autoencoder architecture.
[Figure 5 in paper]
Adding shape feature (f)
for geometric continuity.
[Figure 5 in paper]
Shape features
PM parameters
Example of Light Field Descriptors.
[Chen 2003]
Weblab@The University of Tokyo
Proposed Method: Denoising
● Denoising is applied only to shape features in the results of comparison.
● PM parameters have no hidden process which introduce noise because they fully control the shape.
On the other hand, shape features are sensitive to image-level processing details.
→ Denoising only works on shape features.
12
datasets
Weblab@The University of Tokyo
Evaluation and Results: Setup
● PM rulesets
○ Containers: 72 parameters
○ Trees: 100 parameters
13
Example from Containers.
[Figure 12 in paper]
Example from Trees.
[Figure 10 in paper]
Weblab@The University of Tokyo
Evaluation and Results: Interpolation
● The latent space dimensionality is fixed to 2 or 3 in order to provide intuitive and easy-to-understand interface.
14
Weblab@The University of Tokyo 15
Weblab@The University of Tokyo
Evaluation and Results: Shape Continuity in the Reduced Dimensional Space
A latent space extracted from Container ruleset. It seems shape features are contributing to the shape continuity.
16
PM parameters + shape features PM parameters only
Weblab@The University of Tokyo
Evaluation and Results: Shape Continuity in the Reduced Dimensional Space
A linear interpolation from Trees ruleset.
17
PM params only
PM params +
shape params
Weblab@The University of Tokyo
Evaluation and Results: Shape Continuity in the Reduced Dimensional Space
● Authors evaluate the continuity by introducing a cumulative shape dissimilarity measure, which sums up the
feature difference between neighbors along the axes of the latent space.
● Results show that the dissimilarity decreases in both rulesets compared to the baseline (PM parameters only).
18
S: samples generated in the latent space.
N_i: neighbors of sample i along the axes of
the latent space.
f_i: the feature vector of sample i.
Weblab@The University of Tokyo
Evaluation and Results: User Study
90 design non-expert users are asked to design one favorite container using system #1, then asked to replicate it using
system #2.
Group A: PM params only (#1) → PM params + shape params (#2)
Group B: PM params + shape params (#1) → PM params only (#2)
Users replicated the original container better when using PM params + shape params.
19
PM params only
PM params +
shape params
PM params only
PM params +
shape params
Weblab@The University of Tokyo
Evaluation and Results: User Study
● Users using the proposed system designed and replicated the container faster and also satisfied with the quality
of the replica.
● Users using the proposed system also gave positive feedback on its usability.
20
Weblab@The University of Tokyo
Discussion & Conclusion
Limitation of the proposed method
● Dimensionality reduction can result in making some shapes inaccessible any more which can be achieved by the
original parameters.
○ Trade-off between representational capacity and usability of the interface.
○ “Future work will require more studies in the design space siualization techniques.”
Conclusion
● Introduced autoencoder neural networks for design dimensionality reduction.
● Demonstrated the combination of shape features and PM parameters generates the latent space which is
primarily organized by shape similarity.
● Evaluated that the proposed system improves user experience and productivity.
21
Weblab@The University of Tokyo
References
[Hinton 2006] Geoffrey E Hinton and Ruslan R Salakhutdinov. 2006. Reducing the dimensionality of data with neural networks. Science 313,
5786 (2006), 504–507.
[Yoav 2001] Yoav IH Parish and Pascal Muller. 2001. Procedural modeling of cities. In ACM SIGGRAPH. ACM, 301–308.
[Lintermann 1999] Bernd Lintermann and Oliver Deussen. 1999. Interactive modeling of plants. Computer Graphics and Applications 19, 1
(1999), 56–65.
[Koyama 2014] Yuki Koyama, Daisuke Sakamoto, and Takeo Igarashi. 2014. Crowd-powered parameter analysis for visual design exploration. In
UIST. 65–74.
[Huang 2013] S. Huang, A. Shamir, C. Shen, H. Zhang, A. Sheffer, S. Hu, and D. Cohen-Or. 2013. Qualitative organization of collections of shapes
via quartet analysis. ACM Trans. Graph. 32, 4 (2013), 71.
[Chen 2003] D. Chen, X. Tian, Y. Shen, and M. Ouhyoung. 2003. On visual similarity based 3D model retrieval. In Computer graphics forum, Vol.
22. 223–232.
22

More Related Content

What's hot

A Graph Summarization: A Survey | Summarizing and understanding large graphs
A Graph Summarization: A Survey | Summarizing and understanding large graphsA Graph Summarization: A Survey | Summarizing and understanding large graphs
A Graph Summarization: A Survey | Summarizing and understanding large graphsaftab alam
 
Tablet gaze unconstrained appearance based gaze estimation in mobile tablets
Tablet gaze unconstrained appearance based gaze estimation in mobile tabletsTablet gaze unconstrained appearance based gaze estimation in mobile tablets
Tablet gaze unconstrained appearance based gaze estimation in mobile tabletsJaey Jeong
 
Poster 2D Thinning
Poster 2D ThinningPoster 2D Thinning
Poster 2D ThinningRMwebsite
 
Three different classifiers for facial age estimation based on K-nearest neig...
Three different classifiers for facial age estimation based on K-nearest neig...Three different classifiers for facial age estimation based on K-nearest neig...
Three different classifiers for facial age estimation based on K-nearest neig...Alaa Tharwat
 
Distributed graph summarization
Distributed graph summarizationDistributed graph summarization
Distributed graph summarizationaftab alam
 
Facial Image Analysis for age and gender and
Facial Image Analysis for age and gender andFacial Image Analysis for age and gender and
Facial Image Analysis for age and gender andYuheng Wang
 
Bangla Optical Digits Recognition using Edge Detection Method
Bangla Optical Digits Recognition using Edge Detection MethodBangla Optical Digits Recognition using Edge Detection Method
Bangla Optical Digits Recognition using Edge Detection MethodIOSR Journals
 
Topological Data Analysis and Persistent Homology
Topological Data Analysis and Persistent HomologyTopological Data Analysis and Persistent Homology
Topological Data Analysis and Persistent HomologyCarla Melia
 
Texture descriptor based on local combination adaptive ternary pattern
Texture descriptor based on local combination adaptive ternary patternTexture descriptor based on local combination adaptive ternary pattern
Texture descriptor based on local combination adaptive ternary patternProjectsatbangalore
 
Classification model for predicting student's knowledge
Classification model for predicting student's knowledgeClassification model for predicting student's knowledge
Classification model for predicting student's knowledgeAshish Ranjan
 
IRJET- Object Detection using Hausdorff Distance
IRJET-  	  Object Detection using Hausdorff DistanceIRJET-  	  Object Detection using Hausdorff Distance
IRJET- Object Detection using Hausdorff DistanceIRJET Journal
 
IRJET - Image Classification using CNN
IRJET - Image Classification using CNNIRJET - Image Classification using CNN
IRJET - Image Classification using CNNIRJET Journal
 
New Technique for Image Encryption Based on Choas and Change of MSB
New Technique for Image Encryption Based on Choas and Change of MSBNew Technique for Image Encryption Based on Choas and Change of MSB
New Technique for Image Encryption Based on Choas and Change of MSBEditor IJCATR
 
MATLAB IMPLEMENTATION OF SELF-ORGANIZING MAPS FOR CLUSTERING OF REMOTE SENSIN...
MATLAB IMPLEMENTATION OF SELF-ORGANIZING MAPS FOR CLUSTERING OF REMOTE SENSIN...MATLAB IMPLEMENTATION OF SELF-ORGANIZING MAPS FOR CLUSTERING OF REMOTE SENSIN...
MATLAB IMPLEMENTATION OF SELF-ORGANIZING MAPS FOR CLUSTERING OF REMOTE SENSIN...Daksh Raj Chopra
 

What's hot (20)

A Graph Summarization: A Survey | Summarizing and understanding large graphs
A Graph Summarization: A Survey | Summarizing and understanding large graphsA Graph Summarization: A Survey | Summarizing and understanding large graphs
A Graph Summarization: A Survey | Summarizing and understanding large graphs
 
Tablet gaze unconstrained appearance based gaze estimation in mobile tablets
Tablet gaze unconstrained appearance based gaze estimation in mobile tabletsTablet gaze unconstrained appearance based gaze estimation in mobile tablets
Tablet gaze unconstrained appearance based gaze estimation in mobile tablets
 
Poster 2D Thinning
Poster 2D ThinningPoster 2D Thinning
Poster 2D Thinning
 
Intelligent ebook
Intelligent ebookIntelligent ebook
Intelligent ebook
 
Three different classifiers for facial age estimation based on K-nearest neig...
Three different classifiers for facial age estimation based on K-nearest neig...Three different classifiers for facial age estimation based on K-nearest neig...
Three different classifiers for facial age estimation based on K-nearest neig...
 
Distributed graph summarization
Distributed graph summarizationDistributed graph summarization
Distributed graph summarization
 
Thesis Proposal Presentation
Thesis Proposal PresentationThesis Proposal Presentation
Thesis Proposal Presentation
 
P1121133746
P1121133746P1121133746
P1121133746
 
Facial Image Analysis for age and gender and
Facial Image Analysis for age and gender andFacial Image Analysis for age and gender and
Facial Image Analysis for age and gender and
 
Bangla Optical Digits Recognition using Edge Detection Method
Bangla Optical Digits Recognition using Edge Detection MethodBangla Optical Digits Recognition using Edge Detection Method
Bangla Optical Digits Recognition using Edge Detection Method
 
Topological Data Analysis and Persistent Homology
Topological Data Analysis and Persistent HomologyTopological Data Analysis and Persistent Homology
Topological Data Analysis and Persistent Homology
 
Impact of Graphs and Network in Minimizing Project and Product Cost
Impact of Graphs and Network in Minimizing Project and Product CostImpact of Graphs and Network in Minimizing Project and Product Cost
Impact of Graphs and Network in Minimizing Project and Product Cost
 
CSMR06a.ppt
CSMR06a.pptCSMR06a.ppt
CSMR06a.ppt
 
Texture descriptor based on local combination adaptive ternary pattern
Texture descriptor based on local combination adaptive ternary patternTexture descriptor based on local combination adaptive ternary pattern
Texture descriptor based on local combination adaptive ternary pattern
 
Classification model for predicting student's knowledge
Classification model for predicting student's knowledgeClassification model for predicting student's knowledge
Classification model for predicting student's knowledge
 
IRJET- Object Detection using Hausdorff Distance
IRJET-  	  Object Detection using Hausdorff DistanceIRJET-  	  Object Detection using Hausdorff Distance
IRJET- Object Detection using Hausdorff Distance
 
IRJET - Image Classification using CNN
IRJET - Image Classification using CNNIRJET - Image Classification using CNN
IRJET - Image Classification using CNN
 
New Technique for Image Encryption Based on Choas and Change of MSB
New Technique for Image Encryption Based on Choas and Change of MSBNew Technique for Image Encryption Based on Choas and Change of MSB
New Technique for Image Encryption Based on Choas and Change of MSB
 
MATLAB IMPLEMENTATION OF SELF-ORGANIZING MAPS FOR CLUSTERING OF REMOTE SENSIN...
MATLAB IMPLEMENTATION OF SELF-ORGANIZING MAPS FOR CLUSTERING OF REMOTE SENSIN...MATLAB IMPLEMENTATION OF SELF-ORGANIZING MAPS FOR CLUSTERING OF REMOTE SENSIN...
MATLAB IMPLEMENTATION OF SELF-ORGANIZING MAPS FOR CLUSTERING OF REMOTE SENSIN...
 
Hybrid topology
Hybrid topologyHybrid topology
Hybrid topology
 

Similar to Procedural modeling using autoencoder networks

M tech 1 2 cad cam
M tech 1 2 cad camM tech 1 2 cad cam
M tech 1 2 cad camAKRAMTARIQ2
 
Mlp mixer image_process_210613 deeplearning paper review!
Mlp mixer image_process_210613 deeplearning paper review!Mlp mixer image_process_210613 deeplearning paper review!
Mlp mixer image_process_210613 deeplearning paper review!taeseon ryu
 
An Adjacent Analysis of the Parallel Programming Model Perspective: A Survey
 An Adjacent Analysis of the Parallel Programming Model Perspective: A Survey An Adjacent Analysis of the Parallel Programming Model Perspective: A Survey
An Adjacent Analysis of the Parallel Programming Model Perspective: A SurveyIRJET Journal
 
Parn pyramidal+affine+regression+networks+for+dense+semantic+correspondence
Parn pyramidal+affine+regression+networks+for+dense+semantic+correspondenceParn pyramidal+affine+regression+networks+for+dense+semantic+correspondence
Parn pyramidal+affine+regression+networks+for+dense+semantic+correspondenceNAVER Engineering
 
Implementation of a Simulation Model Using the Systems Dynamics: Case Study f...
Implementation of a Simulation Model Using the Systems Dynamics: Case Study f...Implementation of a Simulation Model Using the Systems Dynamics: Case Study f...
Implementation of a Simulation Model Using the Systems Dynamics: Case Study f...IJERA Editor
 
Machine learning testing survey, landscapes and horizons, the Cliff Notes
Machine learning testing  survey, landscapes and horizons, the Cliff NotesMachine learning testing  survey, landscapes and horizons, the Cliff Notes
Machine learning testing survey, landscapes and horizons, the Cliff NotesHeemeng Foo
 
Image Features Matching and Classification Using Machine Learning
Image Features Matching and Classification Using Machine LearningImage Features Matching and Classification Using Machine Learning
Image Features Matching and Classification Using Machine LearningIRJET Journal
 
Proto Spiral.ppt Proto Spiral.ppt Proto Spiral.ppt Proto Spiral.ppt
Proto Spiral.ppt Proto Spiral.ppt Proto Spiral.ppt Proto Spiral.pptProto Spiral.ppt Proto Spiral.ppt Proto Spiral.ppt Proto Spiral.ppt
Proto Spiral.ppt Proto Spiral.ppt Proto Spiral.ppt Proto Spiral.pptAnirbanBhar3
 
Road Network Extraction using Satellite Imagery.
Road Network Extraction using Satellite Imagery.Road Network Extraction using Satellite Imagery.
Road Network Extraction using Satellite Imagery.SUMITRAJ312049
 
A Survey on Wireless Network Simulators
A Survey on Wireless Network SimulatorsA Survey on Wireless Network Simulators
A Survey on Wireless Network SimulatorsjournalBEEI
 
Interface management incose2014_lisi
Interface management incose2014_lisiInterface management incose2014_lisi
Interface management incose2014_lisiMarco Lisi
 
Analysis of Educational Robotics activities using a machine learning approach
Analysis of Educational Robotics activities using a machine learning approachAnalysis of Educational Robotics activities using a machine learning approach
Analysis of Educational Robotics activities using a machine learning approachLorenzo Cesaretti
 
A Survey of Machine Learning Methods Applied to Computer ...
A Survey of Machine Learning Methods Applied to Computer ...A Survey of Machine Learning Methods Applied to Computer ...
A Survey of Machine Learning Methods Applied to Computer ...butest
 
Performance Evaluation using Blackboard Technique in Software Architecture
Performance Evaluation using Blackboard Technique in Software ArchitecturePerformance Evaluation using Blackboard Technique in Software Architecture
Performance Evaluation using Blackboard Technique in Software ArchitectureEditor IJCATR
 
turecko-150426_pse_01
turecko-150426_pse_01turecko-150426_pse_01
turecko-150426_pse_01Peter Fabo
 
A framework to performance analysis of software architectural styles
A framework to performance analysis of software architectural stylesA framework to performance analysis of software architectural styles
A framework to performance analysis of software architectural stylesijfcstjournal
 
07. MTE - Studi Kasus Pemodelan Sistem.pptx
07. MTE - Studi Kasus Pemodelan Sistem.pptx07. MTE - Studi Kasus Pemodelan Sistem.pptx
07. MTE - Studi Kasus Pemodelan Sistem.pptxAsalReview
 
Self-Directing Text Detection and Removal from Images with Smoothing
Self-Directing Text Detection and Removal from Images with SmoothingSelf-Directing Text Detection and Removal from Images with Smoothing
Self-Directing Text Detection and Removal from Images with SmoothingPriyanka Wagh
 
RuCORD: Rule-based Composite Operation Recovering and Detection to Support Co...
RuCORD: Rule-based Composite Operation Recovering and Detection to Support Co...RuCORD: Rule-based Composite Operation Recovering and Detection to Support Co...
RuCORD: Rule-based Composite Operation Recovering and Detection to Support Co...Amanuel Alemayehu
 

Similar to Procedural modeling using autoencoder networks (20)

M tech 1 2 cad cam
M tech 1 2 cad camM tech 1 2 cad cam
M tech 1 2 cad cam
 
Thesis Giani UIC Slides EN
Thesis Giani UIC Slides ENThesis Giani UIC Slides EN
Thesis Giani UIC Slides EN
 
Mlp mixer image_process_210613 deeplearning paper review!
Mlp mixer image_process_210613 deeplearning paper review!Mlp mixer image_process_210613 deeplearning paper review!
Mlp mixer image_process_210613 deeplearning paper review!
 
An Adjacent Analysis of the Parallel Programming Model Perspective: A Survey
 An Adjacent Analysis of the Parallel Programming Model Perspective: A Survey An Adjacent Analysis of the Parallel Programming Model Perspective: A Survey
An Adjacent Analysis of the Parallel Programming Model Perspective: A Survey
 
Parn pyramidal+affine+regression+networks+for+dense+semantic+correspondence
Parn pyramidal+affine+regression+networks+for+dense+semantic+correspondenceParn pyramidal+affine+regression+networks+for+dense+semantic+correspondence
Parn pyramidal+affine+regression+networks+for+dense+semantic+correspondence
 
Implementation of a Simulation Model Using the Systems Dynamics: Case Study f...
Implementation of a Simulation Model Using the Systems Dynamics: Case Study f...Implementation of a Simulation Model Using the Systems Dynamics: Case Study f...
Implementation of a Simulation Model Using the Systems Dynamics: Case Study f...
 
Machine learning testing survey, landscapes and horizons, the Cliff Notes
Machine learning testing  survey, landscapes and horizons, the Cliff NotesMachine learning testing  survey, landscapes and horizons, the Cliff Notes
Machine learning testing survey, landscapes and horizons, the Cliff Notes
 
Image Features Matching and Classification Using Machine Learning
Image Features Matching and Classification Using Machine LearningImage Features Matching and Classification Using Machine Learning
Image Features Matching and Classification Using Machine Learning
 
Proto Spiral.ppt Proto Spiral.ppt Proto Spiral.ppt Proto Spiral.ppt
Proto Spiral.ppt Proto Spiral.ppt Proto Spiral.ppt Proto Spiral.pptProto Spiral.ppt Proto Spiral.ppt Proto Spiral.ppt Proto Spiral.ppt
Proto Spiral.ppt Proto Spiral.ppt Proto Spiral.ppt Proto Spiral.ppt
 
Road Network Extraction using Satellite Imagery.
Road Network Extraction using Satellite Imagery.Road Network Extraction using Satellite Imagery.
Road Network Extraction using Satellite Imagery.
 
A Survey on Wireless Network Simulators
A Survey on Wireless Network SimulatorsA Survey on Wireless Network Simulators
A Survey on Wireless Network Simulators
 
Interface management incose2014_lisi
Interface management incose2014_lisiInterface management incose2014_lisi
Interface management incose2014_lisi
 
Analysis of Educational Robotics activities using a machine learning approach
Analysis of Educational Robotics activities using a machine learning approachAnalysis of Educational Robotics activities using a machine learning approach
Analysis of Educational Robotics activities using a machine learning approach
 
A Survey of Machine Learning Methods Applied to Computer ...
A Survey of Machine Learning Methods Applied to Computer ...A Survey of Machine Learning Methods Applied to Computer ...
A Survey of Machine Learning Methods Applied to Computer ...
 
Performance Evaluation using Blackboard Technique in Software Architecture
Performance Evaluation using Blackboard Technique in Software ArchitecturePerformance Evaluation using Blackboard Technique in Software Architecture
Performance Evaluation using Blackboard Technique in Software Architecture
 
turecko-150426_pse_01
turecko-150426_pse_01turecko-150426_pse_01
turecko-150426_pse_01
 
A framework to performance analysis of software architectural styles
A framework to performance analysis of software architectural stylesA framework to performance analysis of software architectural styles
A framework to performance analysis of software architectural styles
 
07. MTE - Studi Kasus Pemodelan Sistem.pptx
07. MTE - Studi Kasus Pemodelan Sistem.pptx07. MTE - Studi Kasus Pemodelan Sistem.pptx
07. MTE - Studi Kasus Pemodelan Sistem.pptx
 
Self-Directing Text Detection and Removal from Images with Smoothing
Self-Directing Text Detection and Removal from Images with SmoothingSelf-Directing Text Detection and Removal from Images with Smoothing
Self-Directing Text Detection and Removal from Images with Smoothing
 
RuCORD: Rule-based Composite Operation Recovering and Detection to Support Co...
RuCORD: Rule-based Composite Operation Recovering and Detection to Support Co...RuCORD: Rule-based Composite Operation Recovering and Detection to Support Co...
RuCORD: Rule-based Composite Operation Recovering and Detection to Support Co...
 

More from Shuhei Iitsuka

Online and offline handwritten chinese character recognition a comprehensive...
Online and offline handwritten chinese character recognition  a comprehensive...Online and offline handwritten chinese character recognition  a comprehensive...
Online and offline handwritten chinese character recognition a comprehensive...Shuhei Iitsuka
 
Inferring win–lose product network from user behavior
Inferring win–lose product network from user behaviorInferring win–lose product network from user behavior
Inferring win–lose product network from user behaviorShuhei Iitsuka
 
バリエーションの提示がもたらす長期的効果に着目したウェブサイト最適化手法 @第31回人工知能学会全国大会
バリエーションの提示がもたらす長期的効果に着目したウェブサイト最適化手法 @第31回人工知能学会全国大会バリエーションの提示がもたらす長期的効果に着目したウェブサイト最適化手法 @第31回人工知能学会全国大会
バリエーションの提示がもたらす長期的効果に着目したウェブサイト最適化手法 @第31回人工知能学会全国大会Shuhei Iitsuka
 
Generating sentences from a continuous space
Generating sentences from a continuous spaceGenerating sentences from a continuous space
Generating sentences from a continuous spaceShuhei Iitsuka
 
ウェブサイト最適化のためのバリエーション自動生成システム
ウェブサイト最適化のためのバリエーション自動生成システムウェブサイト最適化のためのバリエーション自動生成システム
ウェブサイト最適化のためのバリエーション自動生成システムShuhei Iitsuka
 
Improving the Sensitivity of Online Controlled Experiments by Utilizing Pre-E...
Improving the Sensitivity of Online Controlled Experiments by Utilizing Pre-E...Improving the Sensitivity of Online Controlled Experiments by Utilizing Pre-E...
Improving the Sensitivity of Online Controlled Experiments by Utilizing Pre-E...Shuhei Iitsuka
 
Machine learning meets web development
Machine learning meets web developmentMachine learning meets web development
Machine learning meets web developmentShuhei Iitsuka
 
Python と Xpath で ウェブからデータをあつめる
Python と Xpath で ウェブからデータをあつめるPython と Xpath で ウェブからデータをあつめる
Python と Xpath で ウェブからデータをあつめるShuhei Iitsuka
 
リミックスからはじめる DTM 入門
リミックスからはじめる DTM 入門リミックスからはじめる DTM 入門
リミックスからはじめる DTM 入門Shuhei Iitsuka
 
【DBDA 勉強会 2013 夏】Chapter 12: Bayesian Approaches to Testing a Point (‘‘Null’’...
【DBDA 勉強会 2013 夏】Chapter 12: Bayesian Approaches to Testing a Point (‘‘Null’’...【DBDA 勉強会 2013 夏】Chapter 12: Bayesian Approaches to Testing a Point (‘‘Null’’...
【DBDA 勉強会 2013 夏】Chapter 12: Bayesian Approaches to Testing a Point (‘‘Null’’...Shuhei Iitsuka
 
Asia Trend Map: Forecasting “Cool Japan” Content Popularity on Web Data
Asia Trend Map: Forecasting “Cool Japan” Content Popularity on Web DataAsia Trend Map: Forecasting “Cool Japan” Content Popularity on Web Data
Asia Trend Map: Forecasting “Cool Japan” Content Popularity on Web DataShuhei Iitsuka
 
【DBDA 勉強会 2013 夏】Doing Bayesian Data Analysis Chapter 4: Bayes’ Rule
【DBDA 勉強会 2013 夏】Doing Bayesian Data Analysis Chapter 4: Bayes’ Rule【DBDA 勉強会 2013 夏】Doing Bayesian Data Analysis Chapter 4: Bayes’ Rule
【DBDA 勉強会 2013 夏】Doing Bayesian Data Analysis Chapter 4: Bayes’ RuleShuhei Iitsuka
 
UT Startup Gym で人生が変わった話
UT Startup Gym で人生が変わった話UT Startup Gym で人生が変わった話
UT Startup Gym で人生が変わった話Shuhei Iitsuka
 
ウェブサイトで収益を得る
ウェブサイトで収益を得るウェブサイトで収益を得る
ウェブサイトで収益を得るShuhei Iitsuka
 
HTML で自己紹介ページをつくる
HTML で自己紹介ページをつくるHTML で自己紹介ページをつくる
HTML で自己紹介ページをつくるShuhei Iitsuka
 
データベースを使おう
データベースを使おうデータベースを使おう
データベースを使おうShuhei Iitsuka
 
ウェブサービスの企画とデザイン
ウェブサービスの企画とデザインウェブサービスの企画とデザイン
ウェブサービスの企画とデザインShuhei Iitsuka
 
データベースを使おう
データベースを使おうデータベースを使おう
データベースを使おうShuhei Iitsuka
 
第3期キックオフ説明会+勉強会
第3期キックオフ説明会+勉強会 第3期キックオフ説明会+勉強会
第3期キックオフ説明会+勉強会 Shuhei Iitsuka
 
かんたん Twitter アプリをつくろう
かんたん Twitter アプリをつくろう かんたん Twitter アプリをつくろう
かんたん Twitter アプリをつくろう Shuhei Iitsuka
 

More from Shuhei Iitsuka (20)

Online and offline handwritten chinese character recognition a comprehensive...
Online and offline handwritten chinese character recognition  a comprehensive...Online and offline handwritten chinese character recognition  a comprehensive...
Online and offline handwritten chinese character recognition a comprehensive...
 
Inferring win–lose product network from user behavior
Inferring win–lose product network from user behaviorInferring win–lose product network from user behavior
Inferring win–lose product network from user behavior
 
バリエーションの提示がもたらす長期的効果に着目したウェブサイト最適化手法 @第31回人工知能学会全国大会
バリエーションの提示がもたらす長期的効果に着目したウェブサイト最適化手法 @第31回人工知能学会全国大会バリエーションの提示がもたらす長期的効果に着目したウェブサイト最適化手法 @第31回人工知能学会全国大会
バリエーションの提示がもたらす長期的効果に着目したウェブサイト最適化手法 @第31回人工知能学会全国大会
 
Generating sentences from a continuous space
Generating sentences from a continuous spaceGenerating sentences from a continuous space
Generating sentences from a continuous space
 
ウェブサイト最適化のためのバリエーション自動生成システム
ウェブサイト最適化のためのバリエーション自動生成システムウェブサイト最適化のためのバリエーション自動生成システム
ウェブサイト最適化のためのバリエーション自動生成システム
 
Improving the Sensitivity of Online Controlled Experiments by Utilizing Pre-E...
Improving the Sensitivity of Online Controlled Experiments by Utilizing Pre-E...Improving the Sensitivity of Online Controlled Experiments by Utilizing Pre-E...
Improving the Sensitivity of Online Controlled Experiments by Utilizing Pre-E...
 
Machine learning meets web development
Machine learning meets web developmentMachine learning meets web development
Machine learning meets web development
 
Python と Xpath で ウェブからデータをあつめる
Python と Xpath で ウェブからデータをあつめるPython と Xpath で ウェブからデータをあつめる
Python と Xpath で ウェブからデータをあつめる
 
リミックスからはじめる DTM 入門
リミックスからはじめる DTM 入門リミックスからはじめる DTM 入門
リミックスからはじめる DTM 入門
 
【DBDA 勉強会 2013 夏】Chapter 12: Bayesian Approaches to Testing a Point (‘‘Null’’...
【DBDA 勉強会 2013 夏】Chapter 12: Bayesian Approaches to Testing a Point (‘‘Null’’...【DBDA 勉強会 2013 夏】Chapter 12: Bayesian Approaches to Testing a Point (‘‘Null’’...
【DBDA 勉強会 2013 夏】Chapter 12: Bayesian Approaches to Testing a Point (‘‘Null’’...
 
Asia Trend Map: Forecasting “Cool Japan” Content Popularity on Web Data
Asia Trend Map: Forecasting “Cool Japan” Content Popularity on Web DataAsia Trend Map: Forecasting “Cool Japan” Content Popularity on Web Data
Asia Trend Map: Forecasting “Cool Japan” Content Popularity on Web Data
 
【DBDA 勉強会 2013 夏】Doing Bayesian Data Analysis Chapter 4: Bayes’ Rule
【DBDA 勉強会 2013 夏】Doing Bayesian Data Analysis Chapter 4: Bayes’ Rule【DBDA 勉強会 2013 夏】Doing Bayesian Data Analysis Chapter 4: Bayes’ Rule
【DBDA 勉強会 2013 夏】Doing Bayesian Data Analysis Chapter 4: Bayes’ Rule
 
UT Startup Gym で人生が変わった話
UT Startup Gym で人生が変わった話UT Startup Gym で人生が変わった話
UT Startup Gym で人生が変わった話
 
ウェブサイトで収益を得る
ウェブサイトで収益を得るウェブサイトで収益を得る
ウェブサイトで収益を得る
 
HTML で自己紹介ページをつくる
HTML で自己紹介ページをつくるHTML で自己紹介ページをつくる
HTML で自己紹介ページをつくる
 
データベースを使おう
データベースを使おうデータベースを使おう
データベースを使おう
 
ウェブサービスの企画とデザイン
ウェブサービスの企画とデザインウェブサービスの企画とデザイン
ウェブサービスの企画とデザイン
 
データベースを使おう
データベースを使おうデータベースを使おう
データベースを使おう
 
第3期キックオフ説明会+勉強会
第3期キックオフ説明会+勉強会 第3期キックオフ説明会+勉強会
第3期キックオフ説明会+勉強会
 
かんたん Twitter アプリをつくろう
かんたん Twitter アプリをつくろう かんたん Twitter アプリをつくろう
かんたん Twitter アプリをつくろう
 

Recently uploaded

Call Girls Narol 7397865700 Independent Call Girls
Call Girls Narol 7397865700 Independent Call GirlsCall Girls Narol 7397865700 Independent Call Girls
Call Girls Narol 7397865700 Independent Call Girlsssuser7cb4ff
 
Instrumentation, measurement and control of bio process parameters ( Temperat...
Instrumentation, measurement and control of bio process parameters ( Temperat...Instrumentation, measurement and control of bio process parameters ( Temperat...
Instrumentation, measurement and control of bio process parameters ( Temperat...121011101441
 
Decoding Kotlin - Your guide to solving the mysterious in Kotlin.pptx
Decoding Kotlin - Your guide to solving the mysterious in Kotlin.pptxDecoding Kotlin - Your guide to solving the mysterious in Kotlin.pptx
Decoding Kotlin - Your guide to solving the mysterious in Kotlin.pptxJoão Esperancinha
 
INFLUENCE OF NANOSILICA ON THE PROPERTIES OF CONCRETE
INFLUENCE OF NANOSILICA ON THE PROPERTIES OF CONCRETEINFLUENCE OF NANOSILICA ON THE PROPERTIES OF CONCRETE
INFLUENCE OF NANOSILICA ON THE PROPERTIES OF CONCRETEroselinkalist12
 
Gurgaon ✡️9711147426✨Call In girls Gurgaon Sector 51 escort service
Gurgaon ✡️9711147426✨Call In girls Gurgaon Sector 51 escort serviceGurgaon ✡️9711147426✨Call In girls Gurgaon Sector 51 escort service
Gurgaon ✡️9711147426✨Call In girls Gurgaon Sector 51 escort servicejennyeacort
 
Arduino_CSE ece ppt for working and principal of arduino.ppt
Arduino_CSE ece ppt for working and principal of arduino.pptArduino_CSE ece ppt for working and principal of arduino.ppt
Arduino_CSE ece ppt for working and principal of arduino.pptSAURABHKUMAR892774
 
main PPT.pptx of girls hostel security using rfid
main PPT.pptx of girls hostel security using rfidmain PPT.pptx of girls hostel security using rfid
main PPT.pptx of girls hostel security using rfidNikhilNagaraju
 
computer application and construction management
computer application and construction managementcomputer application and construction management
computer application and construction managementMariconPadriquez1
 
8251 universal synchronous asynchronous receiver transmitter
8251 universal synchronous asynchronous receiver transmitter8251 universal synchronous asynchronous receiver transmitter
8251 universal synchronous asynchronous receiver transmitterShivangiSharma879191
 
Call Girls Delhi {Jodhpur} 9711199012 high profile service
Call Girls Delhi {Jodhpur} 9711199012 high profile serviceCall Girls Delhi {Jodhpur} 9711199012 high profile service
Call Girls Delhi {Jodhpur} 9711199012 high profile servicerehmti665
 
Past, Present and Future of Generative AI
Past, Present and Future of Generative AIPast, Present and Future of Generative AI
Past, Present and Future of Generative AIabhishek36461
 
Introduction to Machine Learning Unit-3 for II MECH
Introduction to Machine Learning Unit-3 for II MECHIntroduction to Machine Learning Unit-3 for II MECH
Introduction to Machine Learning Unit-3 for II MECHC Sai Kiran
 
Artificial-Intelligence-in-Electronics (K).pptx
Artificial-Intelligence-in-Electronics (K).pptxArtificial-Intelligence-in-Electronics (K).pptx
Artificial-Intelligence-in-Electronics (K).pptxbritheesh05
 
UNIT III ANALOG ELECTRONICS (BASIC ELECTRONICS)
UNIT III ANALOG ELECTRONICS (BASIC ELECTRONICS)UNIT III ANALOG ELECTRONICS (BASIC ELECTRONICS)
UNIT III ANALOG ELECTRONICS (BASIC ELECTRONICS)Dr SOUNDIRARAJ N
 
Electronically Controlled suspensions system .pdf
Electronically Controlled suspensions system .pdfElectronically Controlled suspensions system .pdf
Electronically Controlled suspensions system .pdfme23b1001
 
Study on Air-Water & Water-Water Heat Exchange in a Finned Tube Exchanger
Study on Air-Water & Water-Water Heat Exchange in a Finned Tube ExchangerStudy on Air-Water & Water-Water Heat Exchange in a Finned Tube Exchanger
Study on Air-Water & Water-Water Heat Exchange in a Finned Tube ExchangerAnamika Sarkar
 
An introduction to Semiconductor and its types.pptx
An introduction to Semiconductor and its types.pptxAn introduction to Semiconductor and its types.pptx
An introduction to Semiconductor and its types.pptxPurva Nikam
 

Recently uploaded (20)

Call Girls Narol 7397865700 Independent Call Girls
Call Girls Narol 7397865700 Independent Call GirlsCall Girls Narol 7397865700 Independent Call Girls
Call Girls Narol 7397865700 Independent Call Girls
 
Instrumentation, measurement and control of bio process parameters ( Temperat...
Instrumentation, measurement and control of bio process parameters ( Temperat...Instrumentation, measurement and control of bio process parameters ( Temperat...
Instrumentation, measurement and control of bio process parameters ( Temperat...
 
Decoding Kotlin - Your guide to solving the mysterious in Kotlin.pptx
Decoding Kotlin - Your guide to solving the mysterious in Kotlin.pptxDecoding Kotlin - Your guide to solving the mysterious in Kotlin.pptx
Decoding Kotlin - Your guide to solving the mysterious in Kotlin.pptx
 
INFLUENCE OF NANOSILICA ON THE PROPERTIES OF CONCRETE
INFLUENCE OF NANOSILICA ON THE PROPERTIES OF CONCRETEINFLUENCE OF NANOSILICA ON THE PROPERTIES OF CONCRETE
INFLUENCE OF NANOSILICA ON THE PROPERTIES OF CONCRETE
 
young call girls in Green Park🔝 9953056974 🔝 escort Service
young call girls in Green Park🔝 9953056974 🔝 escort Serviceyoung call girls in Green Park🔝 9953056974 🔝 escort Service
young call girls in Green Park🔝 9953056974 🔝 escort Service
 
Gurgaon ✡️9711147426✨Call In girls Gurgaon Sector 51 escort service
Gurgaon ✡️9711147426✨Call In girls Gurgaon Sector 51 escort serviceGurgaon ✡️9711147426✨Call In girls Gurgaon Sector 51 escort service
Gurgaon ✡️9711147426✨Call In girls Gurgaon Sector 51 escort service
 
Arduino_CSE ece ppt for working and principal of arduino.ppt
Arduino_CSE ece ppt for working and principal of arduino.pptArduino_CSE ece ppt for working and principal of arduino.ppt
Arduino_CSE ece ppt for working and principal of arduino.ppt
 
main PPT.pptx of girls hostel security using rfid
main PPT.pptx of girls hostel security using rfidmain PPT.pptx of girls hostel security using rfid
main PPT.pptx of girls hostel security using rfid
 
computer application and construction management
computer application and construction managementcomputer application and construction management
computer application and construction management
 
8251 universal synchronous asynchronous receiver transmitter
8251 universal synchronous asynchronous receiver transmitter8251 universal synchronous asynchronous receiver transmitter
8251 universal synchronous asynchronous receiver transmitter
 
Call Girls Delhi {Jodhpur} 9711199012 high profile service
Call Girls Delhi {Jodhpur} 9711199012 high profile serviceCall Girls Delhi {Jodhpur} 9711199012 high profile service
Call Girls Delhi {Jodhpur} 9711199012 high profile service
 
Past, Present and Future of Generative AI
Past, Present and Future of Generative AIPast, Present and Future of Generative AI
Past, Present and Future of Generative AI
 
young call girls in Rajiv Chowk🔝 9953056974 🔝 Delhi escort Service
young call girls in Rajiv Chowk🔝 9953056974 🔝 Delhi escort Serviceyoung call girls in Rajiv Chowk🔝 9953056974 🔝 Delhi escort Service
young call girls in Rajiv Chowk🔝 9953056974 🔝 Delhi escort Service
 
Introduction to Machine Learning Unit-3 for II MECH
Introduction to Machine Learning Unit-3 for II MECHIntroduction to Machine Learning Unit-3 for II MECH
Introduction to Machine Learning Unit-3 for II MECH
 
Artificial-Intelligence-in-Electronics (K).pptx
Artificial-Intelligence-in-Electronics (K).pptxArtificial-Intelligence-in-Electronics (K).pptx
Artificial-Intelligence-in-Electronics (K).pptx
 
UNIT III ANALOG ELECTRONICS (BASIC ELECTRONICS)
UNIT III ANALOG ELECTRONICS (BASIC ELECTRONICS)UNIT III ANALOG ELECTRONICS (BASIC ELECTRONICS)
UNIT III ANALOG ELECTRONICS (BASIC ELECTRONICS)
 
Exploring_Network_Security_with_JA3_by_Rakesh Seal.pptx
Exploring_Network_Security_with_JA3_by_Rakesh Seal.pptxExploring_Network_Security_with_JA3_by_Rakesh Seal.pptx
Exploring_Network_Security_with_JA3_by_Rakesh Seal.pptx
 
Electronically Controlled suspensions system .pdf
Electronically Controlled suspensions system .pdfElectronically Controlled suspensions system .pdf
Electronically Controlled suspensions system .pdf
 
Study on Air-Water & Water-Water Heat Exchange in a Finned Tube Exchanger
Study on Air-Water & Water-Water Heat Exchange in a Finned Tube ExchangerStudy on Air-Water & Water-Water Heat Exchange in a Finned Tube Exchanger
Study on Air-Water & Water-Water Heat Exchange in a Finned Tube Exchanger
 
An introduction to Semiconductor and its types.pptx
An introduction to Semiconductor and its types.pptxAn introduction to Semiconductor and its types.pptx
An introduction to Semiconductor and its types.pptx
 

Procedural modeling using autoencoder networks

  • 1. Weblab@The University of Tokyo Procedural Modeling Using Autoencoder Networks Yumer, M. E., Asente, P., Mech, R., and Kara, L. B. Proceedings of the ACM Symposium on User Interface Software and Technology (UIST), 2015. Presenter: Shuhei Iitsuka
  • 2. Weblab@The University of Tokyo Introduction: Problem Procedural modeling (PM) … Geometry modeling created by a set of parameters and algorithms. PM provides rich representation, but it’s difficult to configure a large number of parameters. → More intuitive exploration method is required. Example of PM: Tree modeling with L-system. L-system, Wikipedia Screenshot from a conventional PM interface. [Figure 7 in paper] !?!? 2
  • 3. Weblab@The University of Tokyo Introduction: Contribution Proposed a dimensionality reduction method for PM using autoencoders. An augmentation for continuous latent space generation (without using variational inference!) is also proposed. Contributions: ● A method to create a lower-dimensional representations of PM using autoencoders. ● A PM design system interface using an explore-and-select interaction. 3 Screenshot from the PM design system interface. [Figure 3 in paper]
  • 4. Weblab@The University of Tokyo Why this paper? ● Future extraction should be a key component for generating variations of web design. ● Autoencoders are attracting great attention for future extraction, so I was searching a good application of them in the field of visual design. ● When I queried “visual design” and “autoencoder” on Google Scholar, this paper was the top hit. 4
  • 5. Weblab@The University of Tokyo Agenda 1. Related Works 2. Proposed Method 3. Evaluation and Results 4. Discussion & Conclusion 5
  • 6. Weblab@The University of Tokyo Related Works: Procedural Modeling From when L-system is proposed, procedural modeling (PM) has many applications in digital design. In order to migrate the burden of parameter adjustment, two approaches are proposed. ● Targeted design ● Exploratory system 6 Cities modeled by procedural modeling. [Yoav 2001]
  • 7. Weblab@The University of Tokyo Related Works: Workaround for Parameter Adjustment Targeted Design Streamlines parameter adjustment by providing high-level design grammar. PROS: Users are free from adjusting PM parameters directly CONS: Less user control over the generated models. 7 Exploratory System By providing helpful features for exploration, ease the burden to adjust the parameters (e.g. show frequently used models, devised parameter dialogue) PROS: Users can start modeling from their liking initial model. CONS: Users need to back to the conventional PM approach to tweak the parameters. Design grammar for plant modeling. [Lintermann 1999] Visualization of goodness estimated by using crowdsourcing. [Koyama 2014]
  • 8. Weblab@The University of Tokyo Related Works: Deep Neural Networks and Autoencoders Deep neural networks model abstract characteristics of data using nonlinear transformations. An autoencoder is a special form of a deep neural network which tries to reproduce the input layer at the output layer. Autoencoders is superior to other dimensionality reduction methods [Hinton 2006]. → Employ autoencoders for dimensionality reduction in this research. 8
  • 9. Weblab@The University of Tokyo Proposed Method: User Experience 9 Screenshot from the proposed exploration system interface. [Figure 2 in paper]
  • 10. Weblab@The University of Tokyo Proposed Model: Sampling the Training Data ● Authors employ Categorization tree (C-tree) in order to sample data uniformly with respect to shapes rather than parameters. ○ Create quartets of shaped based on topological structure. ○ Construct the hierarchical tree assembling the quartets. ● After the initial tree is constructed, conduct subsampling As denoted with red rectangles in the figure. ● Iterates (1) adding new sample, (2) computing a new C-tree, (3) performing subsampling until the criteria is satisfied. 10 Example of C-tree. [Figure 4 in paper] Example of topological quartet. [Huang 2013] Close Far
  • 11. Weblab@The University of Tokyo Proposed Method: Neural Network Architecture ● 5-hidden-layer symmetric autoencoder. ● In addition to PM parameters, shape features are included in the input layer in order to ensure geometric continuity. ○ A silhouette-based histogram extracted from Light Field Descriptors is utilized. ● Training is done layer-by-layer followed by a fine tuning of the entire structure. ● Weighted reconstruction error on shape features is applied to the first hidden layer. 11 Autoencoder architecture. [Figure 5 in paper] Adding shape feature (f) for geometric continuity. [Figure 5 in paper] Shape features PM parameters Example of Light Field Descriptors. [Chen 2003]
  • 12. Weblab@The University of Tokyo Proposed Method: Denoising ● Denoising is applied only to shape features in the results of comparison. ● PM parameters have no hidden process which introduce noise because they fully control the shape. On the other hand, shape features are sensitive to image-level processing details. → Denoising only works on shape features. 12 datasets
  • 13. Weblab@The University of Tokyo Evaluation and Results: Setup ● PM rulesets ○ Containers: 72 parameters ○ Trees: 100 parameters 13 Example from Containers. [Figure 12 in paper] Example from Trees. [Figure 10 in paper]
  • 14. Weblab@The University of Tokyo Evaluation and Results: Interpolation ● The latent space dimensionality is fixed to 2 or 3 in order to provide intuitive and easy-to-understand interface. 14
  • 16. Weblab@The University of Tokyo Evaluation and Results: Shape Continuity in the Reduced Dimensional Space A latent space extracted from Container ruleset. It seems shape features are contributing to the shape continuity. 16 PM parameters + shape features PM parameters only
  • 17. Weblab@The University of Tokyo Evaluation and Results: Shape Continuity in the Reduced Dimensional Space A linear interpolation from Trees ruleset. 17 PM params only PM params + shape params
  • 18. Weblab@The University of Tokyo Evaluation and Results: Shape Continuity in the Reduced Dimensional Space ● Authors evaluate the continuity by introducing a cumulative shape dissimilarity measure, which sums up the feature difference between neighbors along the axes of the latent space. ● Results show that the dissimilarity decreases in both rulesets compared to the baseline (PM parameters only). 18 S: samples generated in the latent space. N_i: neighbors of sample i along the axes of the latent space. f_i: the feature vector of sample i.
  • 19. Weblab@The University of Tokyo Evaluation and Results: User Study 90 design non-expert users are asked to design one favorite container using system #1, then asked to replicate it using system #2. Group A: PM params only (#1) → PM params + shape params (#2) Group B: PM params + shape params (#1) → PM params only (#2) Users replicated the original container better when using PM params + shape params. 19 PM params only PM params + shape params PM params only PM params + shape params
  • 20. Weblab@The University of Tokyo Evaluation and Results: User Study ● Users using the proposed system designed and replicated the container faster and also satisfied with the quality of the replica. ● Users using the proposed system also gave positive feedback on its usability. 20
  • 21. Weblab@The University of Tokyo Discussion & Conclusion Limitation of the proposed method ● Dimensionality reduction can result in making some shapes inaccessible any more which can be achieved by the original parameters. ○ Trade-off between representational capacity and usability of the interface. ○ “Future work will require more studies in the design space siualization techniques.” Conclusion ● Introduced autoencoder neural networks for design dimensionality reduction. ● Demonstrated the combination of shape features and PM parameters generates the latent space which is primarily organized by shape similarity. ● Evaluated that the proposed system improves user experience and productivity. 21
  • 22. Weblab@The University of Tokyo References [Hinton 2006] Geoffrey E Hinton and Ruslan R Salakhutdinov. 2006. Reducing the dimensionality of data with neural networks. Science 313, 5786 (2006), 504–507. [Yoav 2001] Yoav IH Parish and Pascal Muller. 2001. Procedural modeling of cities. In ACM SIGGRAPH. ACM, 301–308. [Lintermann 1999] Bernd Lintermann and Oliver Deussen. 1999. Interactive modeling of plants. Computer Graphics and Applications 19, 1 (1999), 56–65. [Koyama 2014] Yuki Koyama, Daisuke Sakamoto, and Takeo Igarashi. 2014. Crowd-powered parameter analysis for visual design exploration. In UIST. 65–74. [Huang 2013] S. Huang, A. Shamir, C. Shen, H. Zhang, A. Sheffer, S. Hu, and D. Cohen-Or. 2013. Qualitative organization of collections of shapes via quartet analysis. ACM Trans. Graph. 32, 4 (2013), 71. [Chen 2003] D. Chen, X. Tian, Y. Shen, and M. Ouhyoung. 2003. On visual similarity based 3D model retrieval. In Computer graphics forum, Vol. 22. 223–232. 22