High-Performance Training of Conditional Random Fields for Large-Scale Applications of Labeling Sequence Data
スポンサーリンク
概要
- 論文の詳細を見る
Conditional random fields (CRFs) have been successfully applied to various applications of predicting and labeling structured data, such as natural language tagging & parsing, image segmentation & object recognition, and protein secondary structure prediction. The key advantages of CRFs are the ability to encode a variety of overlapping, non-independent features from empirical data as well as the capability of reaching the global normalization and optimization. However, estimating parameters for CRFs is very time-consuming due to an intensive forward-backward computation needed to estimate the likelihood function and its gradient during training. This paper presents a high-performance training of CRFs on massively parallel processing systems that allows us to handle huge datasets with hundreds of thousand data sequences and millions of features. We performed the experiments on an important natural language processing task (text chunking) on large-scale corpora and achieved significant results in terms of both the reduction of computational time and the improvement of prediction accuracy.
- 社団法人電子情報通信学会の論文
- 2007-01-01
著者
-
HORIGUCHI Susumu
Graduate School of Information Science, Japan Advanced Institute of Science and Technology (JAIST)
-
Phan Xuan-hieu
Graduate School Of Information Science Japan Advanced Institute Of Science And Technology:(present O
-
Nguyen Le-minh
Graduate School Of Information Science Japan Advanced Institute Of Science And Technology
-
INOGUCHI Yasushi
Graduate School of Information Science, Japan Advanced Institute of Science and Technology
-
Horiguchi Susumu
Tohoku Univ. Sendai‐shi Jpn
-
Horiguchi Susumu
Graduate School Of Information Science Jaist
-
Inoguchi Yasushi
Graduate School Of Information Science Japan Advanced Institute Of Science And Technology
-
Horiguchi Susumu
Graduate School Of Computer Science Japan Advanced Institute Of Science And Technology
関連論文
- Efficient Network Coding-Based Loss Recovery for Reliable Multicast in Wireless Networks
- A More Efficient COPE Architecture for Network Coding in Multihop Wireless Networks
- A Nonblocking Optical Switching Network for Crosstalk-Free Permutation
- Crosstalk-Free Permutation in Photonic Rearrangeable Networks Built on a Combination of Horizontal Expansion and Vertical Stacking of Banyan Networks(Special Issue on Parallel and Distributed Computing, Applications and technologies)
- A Lightpath Restoration Method Using Multi-Backup Paths in WDM Networks
- Dynamic RWA Based on the Combination of Mobile Agents Technique and Genetic Algorithms in WDM Networks with Sparse Wavelength Conversion(Software Agent and Its Applications)
- MOBLE ROBOT LOCALIZATION USIGNG OMINI-DIRECTIONAL VIEW
- TTN : A High Performance Hierarchical Interconnection Network for Massively Parallel Computers
- High-Performance Training of Conditional Random Fields for Large-Scale Applications of Labeling Sequence Data
- Personal Name Resolution Crossover Documents by a Semantics-Based Approach(Natural Language Processing)
- Expected-Credibility-Based Job Scheduling for Reliable Volunteer Computing
- Robust Node Positioning in Wireless Sensor Networks
- Self-Routing Nonblocking WDM Switches Based on Arrayed Waveguide Grating
- Routing Algorithms for Packet/Circuit Switching in Optical Multi-log_2N Networks
- Hybrid Packet-Pheromone-Based Probabilistic Routing for Mobile Ad Hoc Networks
- Fair Scheduling for Delay-Sensitive VoIP Traffic
- Efficient Network Coding-Based Loss Recovery for Reliable Multicast in Wireless Networks
- A More Efficient COPE Architecture for Network Coding in Multihop Wireless Networks
- Breakage prediction-based route maintenance in ad hoc networks (インターネットアーキテクチャ)
- A New Dimension Analysis on Blocking Behavior in Banyan-Based Optical Switching Networks
- Modified Hierarchical 3D-Torus Network
- Dynamic Communication Performance of a Hierarchical Torus Network under Non-uniform Traffic Patterns(Computer Systems)
- New Bounds on the Feedforward Design of Optical Output Buffer Multiplexers and Switches
- Maintaining Packet Order in Reservation-Based Shared-Memory Optical Packet Switch
- Redundant Vias Insertion for Performance Enhancement in 3D ICs
- Efficient routing algorithms for feedforward output buffer queue switch (ネットワークシステム)
- Variant X-Tree Clock Distribution Network and Its Performance Evaluations(Low-Power and High-Performance VLSI Circuit Technology,VLSI Technology toward Frontiers of New Market)
- 無線アドホックネットワークにおける蟻の食性を利用したルーティング法(ユビキタス)
- Parallel Molecular Dynamics in a Parallelizing SML Compiler(Special Issue on Parallel and Distributed Computing, Applications and technologies)
- HTN : A New Hierarchical Interconnection Network for Massively Parallel Computers(Special Issue on Parallel and Distributed Computing, Applications and technologies)
- A Class of Benes-Based Optical Multistage Interconnection Networks for Crosstalk-Free Realization of Permutations(Fiber-Optic Transmission for Communications)
- A more accurate skew model for well-balanced H-tree clock distribution network (プロセス・デバイス・回路シミュレーション(統計モデリングも含む))
- A more accurate skew model for well-balanced H-tree clock distribution network (プロセス・デバイス・回路シミュレーション(統計モデリングも含む))
- Behavior of Active Lightpath Restoration in All-Optical WDM Networks
- An Upper Bound on Blocking Probability for Vertical Stacked Optical Banyan Networks with Extra Stage
- Performance Measurement of the Multi-backup paths Restoration Scheme under Capacity Constraint
- On the Multiple Bridge Fault Diagnosis of Baseline Multistage Interconnection Networks (Special Issue on Architectures, Algorithms and Networks for Massively parallel Computing)
- A Probabilistic Sentence Reduction Using Maximum Entropy Model(Natural Language Processing)
- Multicasting in Multihop Optical WDM Networks with Limited Wavelength Conversion(Special Invited Survey)
- Load Balancing Based on Load Coherence between Continuous Images for an Object-Space Parallel Ray-Tracing System
- Special Issue on Parallel and Distributed Computing, Applications and Technologies