Fisheye
http://bost.ocks.org/mike/fisheye/
D3.js (Data Driven Document)
http://d3js.org/
http://bost.ocks.org/mike/fisheye/
D3.js (Data Driven Document)
http://d3js.org/
gnuplotでeps
14 年前
2014年12月13日土曜日
2014年12月5日金曜日
2014年12月3日水曜日
Deep Learning
Application to Natural Language Processing : http://www.socher.org/
Distributed Neural Networks with GPUs in the AWS Cloud
http://techblog.netflix.com/2014/02/distributed-neural-networks-with-gpus.html
Large Scale Distributed Deep Networks
http://static.googleusercontent.com/media/research.google.com/en//archive/large_deep_networks_nips2012.pdf
Open Software
Deep Learning for Java
http://deeplearning4j.org/
Deeplearning4j is the first commercial-grade, open-source deep-learning library written in Java. It is meant to be used in business environments, rather than as a research tool for extensive data exploration. Deeplearning4j is most helpful in solving distinct problems, like identifying faces, voices, spam or e-commerce fraud. Deeplearning4j integrates with GPUs and includes a versatile n-dimensional array class. DL4J aims to be cutting-edge plug and play, more convention than configuration. By following its conventions, you get an infinitely scalable deep-learning architecture suitable for Hadoop and other big-data structures. This Java deep-learning library has a domain-specific language for neural networks that serves to turn their multiple knobs. Deeplearning4j includes a distributed deep-learning framework and a normal deep-learning framework (i.e. it runs on a single thread as well). Training takes place in the cluster, which means it can process massive amounts of data. Nets are trained in parallel via iterative reduce, and they are equally compatible with Java, Scala and Clojure, since they’re written for the JVM. This open-source, distributed deep-learning framework is made for data input and neural net training at scale, and its output should be highly accurate predictive models. By following the links at the bottom of each page, you will learn to set up, and train with sample data, several types of deep-learning networks. These include single- and multithread networks, Restricted Boltzmann machines, deep-belief networks, Deep Autoencoders, Recursive Neural Tensor Networks, Convolutional Nets and Stacked Denoising Autoencoders.
Distributed Neural Networks with GPUs in the AWS Cloud
http://techblog.netflix.com/2014/02/distributed-neural-networks-with-gpus.html
Large Scale Distributed Deep Networks
http://static.googleusercontent.com/media/research.google.com/en//archive/large_deep_networks_nips2012.pdf
Open Software
Deep Learning for Java
http://deeplearning4j.org/
Deeplearning4j is the first commercial-grade, open-source deep-learning library written in Java. It is meant to be used in business environments, rather than as a research tool for extensive data exploration. Deeplearning4j is most helpful in solving distinct problems, like identifying faces, voices, spam or e-commerce fraud. Deeplearning4j integrates with GPUs and includes a versatile n-dimensional array class. DL4J aims to be cutting-edge plug and play, more convention than configuration. By following its conventions, you get an infinitely scalable deep-learning architecture suitable for Hadoop and other big-data structures. This Java deep-learning library has a domain-specific language for neural networks that serves to turn their multiple knobs. Deeplearning4j includes a distributed deep-learning framework and a normal deep-learning framework (i.e. it runs on a single thread as well). Training takes place in the cluster, which means it can process massive amounts of data. Nets are trained in parallel via iterative reduce, and they are equally compatible with Java, Scala and Clojure, since they’re written for the JVM. This open-source, distributed deep-learning framework is made for data input and neural net training at scale, and its output should be highly accurate predictive models. By following the links at the bottom of each page, you will learn to set up, and train with sample data, several types of deep-learning networks. These include single- and multithread networks, Restricted Boltzmann machines, deep-belief networks, Deep Autoencoders, Recursive Neural Tensor Networks, Convolutional Nets and Stacked Denoising Autoencoders.
2014年1月16日木曜日
渡部君・金刺君とのミーティングメモ
DEIM 2014の論文：Dropbox/SuzumuraLab/Projects/Social Analytics/watanabe/paper/DEIM2014/submit/draft_20140114.pdf
引き継ぎ、研究の方向性
XAXIS（５章）の説明
頂点を追加、削除→オブジェクトを消さず、インアクティブにする
実行時のコスト（ガベージコレクションなど）は小さいが、メモリやノードの消費量が大きくなる
シミュレーションの基盤を成長させていく（成長シミュレーション）
今のプラットフォームのあるべき姿→研究テーマにする
解析：最初のシミュレーション：インアクティブの割合が大きいと効果が大きい
本来は1/10のノードでOK（6〜7千万／4億）
ハードウェア
ノードもダイナミックに追加する（X10ではできない？）
Amazon EC2 のインスタンスを確保して、ストリーム処理する（石井先輩の研究）
VMを必要に応じて予測、確保
ソフトウェア：
インスタンスの必要な量を予測して追加
X10の他に基盤が必要（Hadoopは難しいが、できなくはない→YARNを使用する）
グラフの処理系の上でgiraph上に、ノードを追加、Hadoopを実装
VMやYARNの起動、頂点の追加削除による実行速度が落ちないように、
成長モデルに応じた予測のアルゴリズムモデルを考える
バルク確保によるメモリ消費量とのトレードオフも考慮する
アダプティブなシミュレーション基盤を作る
オーバーヘッドを考えるとXAXISの方が速いと思われるが、渡部先輩の実験結果（参考値）と比較する
SocialMediaのデータ解析は、シミュレーションのモデルを作る上で重要
スキルトランスファーをしてもらう
確実な方法があるのか、試行錯誤する部分もあるのか
まずは、HadoopをTSUBAMEで動かして、使い方を学ぶ
解析をしたい時にRAがオーバーヘッドなく扱えるよう、ドキュメントを作っておく
何を調べたいか、やりたいことは色々あるが、データが大きいため扱いにくい
一部のデータ（属性情報）だけ取ってきて、それだけでフォローのモデルを作成する
データ同化にもつながる（シミュレーションしながらモデルを改良していく）
新規のユーザ追加のコストを削減
いきなり数十ノードを最初から確保するのは現実的でない
まずは小規模で実行し、ユーザ数が増えてきたら結果をダンプする
規模が大きくなったら途中結果を読み込んで、やり直す
スナップショットを少しずつ取りながら、ノードを確保していく
X10は動的なアロケーションができない
他の新規ノードが追加できるような基盤を作る
twitterの追加解析も行う
クラスタリング係数値、Connected Components
ScaleGraphのライブラリも使っていく
Webマイニングの話にもかかわる
既存研究のWebマイニングの論文を調べて、どうネットワーク解析をしていくか
デフォルトの（通常行われている）解析セットを調査する
そこから発展させて、分かっていないことを仮説を立てて検証する
「何次の隔たり」よりも、もう少しミクロ（詳細）な仮説の方が良い
tweet そのものの解析も行う
投稿した内容、頻度から、ユーザのプロパティを予想する
データ取得（クロール）の方法も教えてもらう
“What is twitter”で行われていないことは？
retweetの木構造も研究されていた
リアルタイムに変わるGraphは、PageRankを計算すると異なる
グラフの構造の変化はリアルタイムにある
グラフ分割、BC、PageRank、クラスタリングをグラフストリーム的に計算する
tweetも、テキスト処理をしてクラスタリング（類似度計算）できる
引き継ぎ、研究の方向性
XAXIS（５章）の説明
頂点を追加、削除→オブジェクトを消さず、インアクティブにする
実行時のコスト（ガベージコレクションなど）は小さいが、メモリやノードの消費量が大きくなる
シミュレーションの基盤を成長させていく（成長シミュレーション）
今のプラットフォームのあるべき姿→研究テーマにする
解析：最初のシミュレーション：インアクティブの割合が大きいと効果が大きい
本来は1/10のノードでOK（6〜7千万／4億）
ハードウェア
ノードもダイナミックに追加する（X10ではできない？）
Amazon EC2 のインスタンスを確保して、ストリーム処理する（石井先輩の研究）
VMを必要に応じて予測、確保
ソフトウェア：
インスタンスの必要な量を予測して追加
X10の他に基盤が必要（Hadoopは難しいが、できなくはない→YARNを使用する）
グラフの処理系の上でgiraph上に、ノードを追加、Hadoopを実装
VMやYARNの起動、頂点の追加削除による実行速度が落ちないように、
成長モデルに応じた予測のアルゴリズムモデルを考える
バルク確保によるメモリ消費量とのトレードオフも考慮する
アダプティブなシミュレーション基盤を作る
オーバーヘッドを考えるとXAXISの方が速いと思われるが、渡部先輩の実験結果（参考値）と比較する
SocialMediaのデータ解析は、シミュレーションのモデルを作る上で重要
スキルトランスファーをしてもらう
確実な方法があるのか、試行錯誤する部分もあるのか
まずは、HadoopをTSUBAMEで動かして、使い方を学ぶ
解析をしたい時にRAがオーバーヘッドなく扱えるよう、ドキュメントを作っておく
何を調べたいか、やりたいことは色々あるが、データが大きいため扱いにくい
一部のデータ（属性情報）だけ取ってきて、それだけでフォローのモデルを作成する
データ同化にもつながる（シミュレーションしながらモデルを改良していく）
新規のユーザ追加のコストを削減
いきなり数十ノードを最初から確保するのは現実的でない
まずは小規模で実行し、ユーザ数が増えてきたら結果をダンプする
規模が大きくなったら途中結果を読み込んで、やり直す
スナップショットを少しずつ取りながら、ノードを確保していく
X10は動的なアロケーションができない
他の新規ノードが追加できるような基盤を作る
twitterの追加解析も行う
クラスタリング係数値、Connected Components
ScaleGraphのライブラリも使っていく
Webマイニングの話にもかかわる
既存研究のWebマイニングの論文を調べて、どうネットワーク解析をしていくか
デフォルトの（通常行われている）解析セットを調査する
そこから発展させて、分かっていないことを仮説を立てて検証する
「何次の隔たり」よりも、もう少しミクロ（詳細）な仮説の方が良い
tweet そのものの解析も行う
投稿した内容、頻度から、ユーザのプロパティを予想する
データ取得（クロール）の方法も教えてもらう
“What is twitter”で行われていないことは？
retweetの木構造も研究されていた
リアルタイムに変わるGraphは、PageRankを計算すると異なる
グラフの構造の変化はリアルタイムにある
グラフ分割、BC、PageRank、クラスタリングをグラフストリーム的に計算する
tweetも、テキスト処理をしてクラスタリング（類似度計算）できる
2014年1月10日金曜日
ACM SIGSIM PADS 2013
Here is a list of accepted papers for PADS 2013 that are all relevant to our agent-based simulations.
http://www.acm-sigsim-pads.org/acceptedPapers.htm
http://www.acm-sigsim-pads.org/acceptedPapers.htm
- Accelerating Optimistic HLA-based Simulations in Virtual Execution Environments
- A Time Management Optimization Framework for Large-Scale Distributed Hardware-In-The-Loop Simulation
- An Expansion-aided Synchronous Conservative Time Management Algorithm on GPU
- Towards Performance Evaluation of Conservative Distributed Discrete-Event Network Simulations Using Second-Order Simulation
- Event Pool Structures for PDES on Many-Core Beowulf Clusters
- Modeling Communication Software Execution for Accurate Simulation of Distributed Systems
- GPU Accelerated Three-stage Execution Model for Event-Parallel Simulation
- Designing Computational Steering Facilities for Distributed Agent Based Simulations
- TerraME HPA: Parallel Simulation of Multi-Agent Systems over SMPs
- Optimizing Parallel Simulation of Multicore Systems Using Domain-Specific Knowledge
- Data Assimilation in Agent Based Simulation of Smart Environment
- On the Parallel Simulation of Scale-Free Networks
- Warp Speed: Executing Time Warp on 1,966,080 Cores
GraphLab's Blog
A co-founder of a fast-growing startup company named "Graphlab" - launched from Carnegie Mellon University - has just blogged about our ScaleGraph project. That's a great news for celebrating a new year 2014, and we definitely accelerate the speed to become the leading project in this area.
http://bickson.blogspot.ie/
http://bickson.blogspot.ie/
2014年1月7日火曜日
2014年1月2日木曜日
Distributed-Memory Parallel Algorithms for Generating Massive Scale-free Networks Using Preferential Attachment Model
Distributed-Memory Parallel Algorithms for Generating Massive Scale-free Networks Using Preferential Attachment Model
2013年12月27日金曜日
TurboGraph
TurboGraph: a fast parallel graph engine handling billion-scale graphs in a single PC
http://dl.acm.org/citation.cfm?id=2487581
Graphs are used to model many real objects such as social networks and web graphs. Many real applications in various fields require efficient and effective management of large-scale graph structured data. Although distributed graph engines such as GBase and Pregel handle billion-scale graphs, the user needs to be skilled at managing and tuning a distributed system in a cluster, which is a nontrivial job for the ordinary user. Furthermore, these distributed systems need many machines in a cluster in order to provide reasonable performance. In order to address this problem, a disk-based parallel graph engine called Graph-Chi, has been recently proposed. Although Graph-Chi significantly outperforms all representative (disk-based) distributed graph engines, we observe that Graph-Chi still has serious performance problems for many important types of graph queries due to 1) limited parallelism and 2) separate steps for I/O processing and CPU processing. In this paper, we propose a general, disk-based graph engine called TurboGraph to process billion-scale graphs very efficiently by using modern hardware on a single PC. TurboGraph is the first truly parallel graph engine that exploits 1) full parallelism including multi-core parallelism and FlashSSD IO parallelism and 2) full overlap of CPU processing and I/O processing as much as possible. Specifically, we propose a novel parallel execution model, calledpin-and-slide. TurboGraph also provides engine-level operators such as BFS which are implemented under the pin-and-slide model. Extensive experimental results with large real datasets show that TurboGraph consistently and significantly outperforms Graph-Chi by up to four orders of magnitude! Our implementation of TurboGraph is available at ``http://wshan.net/turbograph}" as executable files.
Graphs are used to model many real objects such as social networks and web graphs. Many real applications in various fields require efficient and effective management of large-scale graph structured data. Although distributed graph engines such as GBase and Pregel handle billion-scale graphs, the user needs to be skilled at managing and tuning a distributed system in a cluster, which is a nontrivial job for the ordinary user. Furthermore, these distributed systems need many machines in a cluster in order to provide reasonable performance. In order to address this problem, a disk-based parallel graph engine called Graph-Chi, has been recently proposed. Although Graph-Chi significantly outperforms all representative (disk-based) distributed graph engines, we observe that Graph-Chi still has serious performance problems for many important types of graph queries due to 1) limited parallelism and 2) separate steps for I/O processing and CPU processing. In this paper, we propose a general, disk-based graph engine called TurboGraph to process billion-scale graphs very efficiently by using modern hardware on a single PC. TurboGraph is the first truly parallel graph engine that exploits 1) full parallelism including multi-core parallelism and FlashSSD IO parallelism and 2) full overlap of CPU processing and I/O processing as much as possible. Specifically, we propose a novel parallel execution model, calledpin-and-slide. TurboGraph also provides engine-level operators such as BFS which are implemented under the pin-and-slide model. Extensive experimental results with large real datasets show that TurboGraph consistently and significantly outperforms Graph-Chi by up to four orders of magnitude! Our implementation of TurboGraph is available at ``http://wshan.net/turbograph}" as executable files.
2013年12月2日月曜日
System Software Papers in ICWS 2012
Here is a list of papers on system softwares in ICWS 2012. A keyword "service" is widely used in this conference, so important thing is how our efforts could be abstracted out towards the combination with "Service".
- Highly Resilient Systems for Cloud
- Parallel Computing Framework as a Cloud Service
- Overcoming Large Data Transfer Bottlenecks in RESTful Service Orchestrations, ICWS 2012
- RESTful Web Service Mashup Based Coal Mine Safety Monitoring and Control Automation with Wireless Sensor Network, , ICWS 2012
- Intelligent Database Placement in Cloud Environment, , ICWS 2012
- Andes: A Highly Scalable Persistent Messaging System , ICWS 2012
- Enabling Advanced Loading Strategies for Data Intensive Web Services, ICWS 2012
- Green Web Services: Modeling and Estimating Power Consumption of Web Services, ICWS 2012
- A Network Coordinate Based Web Service Positioning Framework for Response Time Prediction, ICWS 2012
- Disk-Offload Middleware for Web-Services Using the Application-Caching Paradigm, ICWS 2012
- Scaling Spatial Alarm Services on Road Networks, ICWS 2012
2013年12月1日日曜日
ICWS 2014
I have been invited as a program committee member for ICWS 2014. Over the past several years, I was involved with some projects related to service oriented computing by focusing more on system software level optimization methods such as differential parsing/deserialization method for high performance XML processing. By just reading through a series of titles in ICWS 2013, most of the researches do not really sound attractive in a practical setting in real world. They still find out a way on how to formulate composite web services in an automatic manner, which is a long standing problem and never solved. That's not really what we should pursue in this research area.
However we can not really ignore this conference since it is treated as a top-tier conference. So what do we do in this context ? We have been pursing on big data / graph processing, stream computing, large-scale agent simulation technology, and so forth. Sometimes you can find a good research topic by looking at the boarder space between two different research areas. In that way, we can start new research topic. So what are they ?
However we can not really ignore this conference since it is treated as a top-tier conference. So what do we do in this context ? We have been pursing on big data / graph processing, stream computing, large-scale agent simulation technology, and so forth. Sometimes you can find a good research topic by looking at the boarder space between two different research areas. In that way, we can start new research topic. So what are they ?
2013年7月12日金曜日
2013年7月9日火曜日
道路ネットワーク縮退化による高速化
金刺君の実験により、Open Street Map の冗長な道路ネットワークから、必要のない交差点を統合すると、かなり高速化できることが確認できた。現状はナイーブな方法（一本道は統合）だが、精度を加味して、動的に統合することも面白いだろう。単に統合するのではなく、仮想的な道路を作るのでも良い。また統合することにより、車両の位置（緯度、経度）が不正確になるが、これはマッピング情報を作るべき。
2013年6月6日木曜日
Facebook's TAO
TAO: Facebook’s Distributed Data Store for the Social Graph
We introduce a simple data model and API tailored for serving the social graph, and TAO, an implementation of this model. TAO is a geographically distributed data store that provides efficient and timely access to the social graph for Facebook’s demanding workload using a fixed set of queries. It is deployed at Facebook, replacing memcache for many data types that fit its model. The system runs on thousands of machines, is widely distributed, and provides access to many petabytes of data. TAO can process a billion reads and millions of writes each second
https://www.usenix.org/conference/atc13/tao-facebook%E2%80%99s-distributed-data-store-social-graph
We introduce a simple data model and API tailored for serving the social graph, and TAO, an implementation of this model. TAO is a geographically distributed data store that provides efficient and timely access to the social graph for Facebook’s demanding workload using a fixed set of queries. It is deployed at Facebook, replacing memcache for many data types that fit its model. The system runs on thousands of machines, is widely distributed, and provides access to many petabytes of data. TAO can process a billion reads and millions of writes each second
https://www.usenix.org/conference/atc13/tao-facebook%E2%80%99s-distributed-data-store-social-graph
2013年5月30日木曜日
Graph Applications in Cybersecurity Domain
Large Scale Graph Analytics and Randomized Algorithms for Applications
in Cybersecurity
http://dl.acm.org/citation.cfm?id=2459984
http://www.graphanalysis.org/SIAM-CSE13/02_Johnson.pdf
A network hacking attack in which hackers repeatedly steal password hashes and move through a computer network with the goal of reaching a computer with high level administrative privileges is known as a pass-the-hash attack. In this paper we apply graph coarsening on graphs obtained from computer network data for the purpose of (a) detecting hackers using this attack and (b) assessing the risk level of the network's current state. We repeatedly contract edges (obtaining a graph minor), which preserves the existence of paths in the graph, and take powers of the adjacency matrix to count the paths. This allows us to detect the existence of paths as well as find paths that have high risk of being exploited by adversaries.
W. Eberle and L. Holder. Applying graph-based anomaly detection approaches to the discovery of insider threats. In IEEE International Conference on Intelligence and Security Informatics (ISI), 2009.
D. A. Spielman and N. Srivastava. Graph sparsification by effective resistances. In Proc. 40th Annual ACM Symposium on Theory of Computing, 2008.
S. Jajodia, S. Noel, and B. O'Berry. Topological analysis of network attack vulnerability. In V. Kumar, J. Srivastava, and A. Lazarevic, editors, Managing Cyber Threats: Issues, Approaches and Challenges, pages 248--266. Kluwer Academic Publisher, 2005.
http://vulcan.ee.iastate.edu/~gmani/personal/papers/journals/IEEE-PS-08.pdf
Measuring Security Risk of Networks Using Attack Graphs
http://users.encs.concordia.ca/~wang/papers/ijngc10.pdf
U Kang's Research Goal Statement
http://www.cs.cmu.edu/~ukang/ukang-research.pdf
http://www.eecs.wsu.edu/~holder/pubs/EberleCATCH09.pdf
in Cybersecurity
http://dl.acm.org/citation.cfm?id=2459984
http://www.graphanalysis.org/SIAM-CSE13/02_Johnson.pdf
A network hacking attack in which hackers repeatedly steal password hashes and move through a computer network with the goal of reaching a computer with high level administrative privileges is known as a pass-the-hash attack. In this paper we apply graph coarsening on graphs obtained from computer network data for the purpose of (a) detecting hackers using this attack and (b) assessing the risk level of the network's current state. We repeatedly contract edges (obtaining a graph minor), which preserves the existence of paths in the graph, and take powers of the adjacency matrix to count the paths. This allows us to detect the existence of paths as well as find paths that have high risk of being exploited by adversaries.
W. Eberle and L. Holder. Applying graph-based anomaly detection approaches to the discovery of insider threats. In IEEE International Conference on Intelligence and Security Informatics (ISI), 2009.
D. A. Spielman and N. Srivastava. Graph sparsification by effective resistances. In Proc. 40th Annual ACM Symposium on Theory of Computing, 2008.
S. Jajodia, S. Noel, and B. O'Berry. Topological analysis of network attack vulnerability. In V. Kumar, J. Srivastava, and A. Lazarevic, editors, Managing Cyber Threats: Issues, Approaches and Challenges, pages 248--266. Kluwer Academic Publisher, 2005.
http://vulcan.ee.iastate.edu/~gmani/personal/papers/journals/IEEE-PS-08.pdf
Measuring Security Risk of Networks Using Attack Graphs
http://users.encs.concordia.ca/~wang/papers/ijngc10.pdf
U Kang's Research Goal Statement
http://www.cs.cmu.edu/~ukang/ukang-research.pdf
http://www.eecs.wsu.edu/~holder/pubs/EberleCATCH09.pdf
2013年5月22日水曜日
Cassovary: Twitter's Graph Processing Library
http://engineering.twitter.com/2012/03/cassovary-big-graph-processing-library.html
2013年4月21日日曜日
2013年3月29日金曜日
Open Data
Nokia Mobile Data Challenge
http://research.nokia.com/page/12000
http://publications.idiap.ch/downloads/papers/2012/Laurila_MDC_2012.pdf
Orange D4D Challenge
http://www.d4d.orange.com/
Open Data - Nottingham
http://www.opendatanottingham.org.uk/catalogue.aspx
Kenya Open Data
http://www.princeton.edu/successfulsocieties/content/data/policy_note/PN_id206/Policy_Note_ID206.pdf
Dublinked
http://www.dublinked.ie/
FutureICT
http://www.futurict.eu/
http://research.nokia.com/page/12000
http://publications.idiap.ch/downloads/papers/2012/Laurila_MDC_2012.pdf
Orange D4D Challenge
http://www.d4d.orange.com/
Open Data - Nottingham
http://www.opendatanottingham.org.uk/catalogue.aspx
Kenya Open Data
http://www.princeton.edu/successfulsocieties/content/data/policy_note/PN_id206/Policy_Note_ID206.pdf
Dublinked
http://www.dublinked.ie/
FutureICT
http://www.futurict.eu/
2012年11月9日金曜日
Location Prediction from Tweets
http://infolab.cse.tamu.edu/static/papers/cikm1184c-cheng.pdf
http://www.hpl.hp.com/research/scl/papers/socialmedia/socialmedia.pdf
http://doras.dcu.ie/16754/1/nohare_paper.pdf
http://www.hpl.hp.com/research/scl/papers/socialmedia/socialmedia.pdf
http://doras.dcu.ie/16754/1/nohare_paper.pdf
2012年11月6日火曜日
2012年9月17日月曜日
Spectral Analysis for Billion-Scale Graphs: Discoveries and Implementation
Spectral Analysis for Billion-Scale Graphs: Discoveries and Implementation
http://www.cs.cmu.edu/~ukang/papers/HeigenPAKDD2011.pdf
Ogata-kun's work should refer the above effort ...
http://www.cs.cmu.edu/~ukang/papers/HeigenPAKDD2011.pdf
Ogata-kun's work should refer the above effort ...
2012年8月13日月曜日
Origin-Destination Data Generation
Deriving origin–destination data from a mobile phone network
http://www.esi2.us.es/GT/docs/iet_art1.pdf
DYNAMIC ORIGIN-DESTINATION MATRIX ESTIMATION AND PREDICTION FOR REAL- TIME TRAFFIC MANAGEMENT SYSTEMS
http://trid.trb.org/view.aspx?id=640422
Time-Varying Network Tomography : Router Link Data
ftp://cs.bell-labs.com/cm/stat/doc/NetTomography.pdf
Development of Revised Methodology for Collecting Origin-Destination Data
http://www.dot.state.fl.us/research-center/Completed_Proj/Summary_PL/FDOT_BD544_30_rpt.pdf
Real-Time Estimation of Origin-Destination Matrices with Partial Trajectories from Electronic Toll Collection Tag Data
http://pubsindex.trb.org/view.aspx?id=775157
Dynamic Origin–Destination Demand Estimation Using Automatic Vehicle Identification Data
http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=1603556&userType=inst
http://www.esi2.us.es/GT/docs/iet_art1.pdf
DYNAMIC ORIGIN-DESTINATION MATRIX ESTIMATION AND PREDICTION FOR REAL- TIME TRAFFIC MANAGEMENT SYSTEMS
http://trid.trb.org/view.aspx?id=640422
Time-Varying Network Tomography : Router Link Data
ftp://cs.bell-labs.com/cm/stat/doc/NetTomography.pdf
Development of Revised Methodology for Collecting Origin-Destination Data
http://www.dot.state.fl.us/research-center/Completed_Proj/Summary_PL/FDOT_BD544_30_rpt.pdf
Real-Time Estimation of Origin-Destination Matrices with Partial Trajectories from Electronic Toll Collection Tag Data
http://pubsindex.trb.org/view.aspx?id=775157
Dynamic Origin–Destination Demand Estimation Using Automatic Vehicle Identification Data
http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=1603556&userType=inst
Distributed Approach for Estimation of Dynamic Origin—Destination Demand
http://trb.metapress.com/content/96400l51428r4737/2012年8月10日金曜日
2012年8月9日木曜日
Gephi
Gephi (http://gephi.org/)
Gephi is an open-source and multiplatform software distributed under the dual license CDDL 1.0 and GNU General Public License v3.
Gephi is an open-source and multiplatform software distributed under the dual license CDDL 1.0 and GNU General Public License v3.
Layout algorithms give the shape to the graph. Gephi provides state-of-the-art algorithms layout algorithms, both for efficiency and quality. The Layout palette allows user to change layout settings while running, and therefore dramatically increase user feedback and experience: Force-based algorithms or Multi-level algorithms (graph coarsening).
Video demonstration :
https://gephi.org/videos/
Time-evolving network visualization
Hashikawa-kun will be working on the time-evolving twitter network since 2006.
The reference demo might be as follows :
Egyptian Revolution on Twitter
https://gephi.org/tag/dynamics/
http://www.youtube.com/watch?v=9guRNGiNBAg
http://www.youtube.com/watch?v=KKNxulf7RNg&feature=relmfu
http://www.youtube.com/watch?v=Hos4m3iO108&feature=related
http://www.youtube.com/watch?v=hI-7wteRv7E
http://computationallegalstudies.com/2010/01/06/the-time-evolving-structure-of-the-gawaher-islamic-forum-as-experienced-by-umar-farouk-abdulmutallab-the-christmas-day-bomber/
http://www.youtube.com/watch?v=a7uTKwbsFtg&feature=related
Not relevant but still interesting visualization:
http://www.evolutionoftheweb.com/?hl=ja#/growth/day
The reference demo might be as follows :
Egyptian Revolution on Twitter
https://gephi.org/tag/dynamics/
http://www.youtube.com/watch?v=9guRNGiNBAg
http://www.youtube.com/watch?v=KKNxulf7RNg&feature=relmfu
http://www.youtube.com/watch?v=Hos4m3iO108&feature=related
http://www.youtube.com/watch?v=hI-7wteRv7E
http://computationallegalstudies.com/2010/01/06/the-time-evolving-structure-of-the-gawaher-islamic-forum-as-experienced-by-umar-farouk-abdulmutallab-the-christmas-day-bomber/
http://www.youtube.com/watch?v=a7uTKwbsFtg&feature=related
Not relevant but still interesting visualization:
http://www.evolutionoftheweb.com/?hl=ja#/growth/day
2012年8月2日木曜日
Research on Microblog services / Twitter
What is Twitter, a Social Network or a News Media?http://an.kaist.ac.kr/~haewoon/papers/2010-www-twitter.pdf
A geographic study of tie strength in social media
http://dl.acm.org/citation.cfm?id=2063959&CFID=100588297&CFTOKEN=89581781
Outtweeting the Twitterers - Predicting Information Cascades in Microblogs
http://static.usenix.org/event/wosn10/tech/full_papers/Galuba.pdf
Why Rumors Spread Fast in Social Networks
http://www.mpi-inf.mpg.de/~tfried/paper/2012CACM.pdf
Measuring Behavioral Trust in Social Network
http://www.cs.rpi.edu/research/pdf/10-03.pdf
Time-Based Sampling of Social Network Activity Graphs
http://www.cs.purdue.edu/homes/neville/papers/ahmed-mlg2010.pdf
Massive Social Network Analysis: Mining Twitter for Social Good
http://www.cc.gatech.edu/~bader/papers/MassiveTwitter-ICPP2010.pdf
Understanding Latent Interactions in Online Social Networks
http://cs.ucsb.edu/~Ravenben/publications/pdf/renren-imc10.pdf
A geographic study of tie strength in social media
http://dl.acm.org/citation.cfm?id=2063959&CFID=100588297&CFTOKEN=89581781
Outtweeting the Twitterers - Predicting Information Cascades in Microblogs
http://static.usenix.org/event/wosn10/tech/full_papers/Galuba.pdf
Why Rumors Spread Fast in Social Networks
http://www.mpi-inf.mpg.de/~tfried/paper/2012CACM.pdf
Measuring Behavioral Trust in Social Network
http://www.cs.rpi.edu/research/pdf/10-03.pdf
Time-Based Sampling of Social Network Activity Graphs
http://www.cs.purdue.edu/homes/neville/papers/ahmed-mlg2010.pdf
Massive Social Network Analysis: Mining Twitter for Social Good
http://www.cc.gatech.edu/~bader/papers/MassiveTwitter-ICPP2010.pdf
Understanding Latent Interactions in Online Social Networks
http://cs.ucsb.edu/~Ravenben/publications/pdf/renren-imc10.pdf
As I introduced it a bit at yesterday's seminar, the crawled data set for Facebook is downloadable from the following URL.
http://suzumura-lab.blogspot.jp/2010/12/streamgraph-facebook-data.html
http://suzumura-lab.blogspot.jp/2010/12/streamgraph-facebook-data.html
2012年7月11日水曜日
Status
雁瀬君 -> 坂本君が作っている X10 ベースの GIM-V 処理系をベースに スケーラブルなIncremental GIM-V 処理系を作る
上野君 -> general model for graph processing, SSSP for 8/8
渡部君 -> Giraph on TSUBAME, 性能分析
岡田君-> 感情分類器を作成中
小形君 -> ParGraph2012 の論文に着手
Charuwat 君 -> RDF data に対するグラフ解析の検討。BCを ParGraphに出す
橋川君->Twitter REST API を用いてクローラ開発
上野君 -> general model for graph processing, SSSP for 8/8
渡部君 -> Giraph on TSUBAME, 性能分析
岡田君-> 感情分類器を作成中
小形君 -> ParGraph2012 の論文に着手
Charuwat 君 -> RDF data に対するグラフ解析の検討。BCを ParGraphに出す
橋川君->Twitter REST API を用いてクローラ開発
登録:
投稿 (Atom)
