Filters








6 Hits in 4.3 sec

TF Boosted Trees: A scalable TensorFlow based framework for gradient boosting [article]

Natalia Ponomareva, Soroush Radpour, Gilbert Hendry, Salem Haykal, Thomas Colthurst, Petr Mitrichev, Alexander Grushetsky
2017 arXiv   pre-print
TF Boosted Trees (TFBT) is a new open-sourced frame-work for the distributed training of gradient boosted trees.  ...  It is based on TensorFlow, and its distinguishing features include a novel architecture, automatic loss differentiation, layer-by-layer boosting that results in smaller ensembles and faster prediction,  ...  server (PS) approach (similar to TencentBoost [4] and PSMART [10] ) is applied, where each worker and PS aggregates statistics only for a subset of features.  ... 
arXiv:1710.11555v1 fatcat:iuef46jawnh7zh5yuxpuaqwtby

InfDetect: a Large Scale Graph-based Fraud Detection System for E-Commerce Insurance [article]

Cen Chen, Chen Liang, Jianbin Lin, Li Wang, Ziqi Liu, Xinxing Yang, Xiukun Wang, Jun Zhou, Yang Shuang, Yuan Qi
2020 arXiv   pre-print
Fraudulent claims towards online insurance typically involve multiple parties such as buyers, sellers, and express companies, and they could lead to heavy financial losses.  ...  Cases on widely applied e-commerce insurance are described to demonstrate the usage and capability of our system.  ...  Our parameter server based GBDT method-PSMART [19] is used as the base classification model. Grid search is performed to find the best parameter settings.  ... 
arXiv:2003.02833v3 fatcat:3ov6jodf2zekjck2f54lanko6m

Compact Multi-Class Boosted Trees [article]

Natalia Ponomareva, Thomas Colthurst, Gilbert Hendry, Salem Haykal, Soroush Radpour
2017 arXiv   pre-print
The first improvement extends the boosting formalism from scalar-valued trees to vector-valued trees.  ...  This allows individual trees to be used as multiclass classifiers, rather than requiring one tree per class, and drastically reduces the model size required for multiclass problems.  ...  In order to handle vector regression or multiclass classification problems, multiple scalar-leaved trees must be used.  ... 
arXiv:1710.11547v1 fatcat:e23td22nwvdmpl5e2jwfz6etva

FunnyBase: a systems level functional annotation of Fundulus ESTs for the analysis of gene expression

Justin E Paschall, Marjorie F Oleksiak, Jeffrey D VanWye, Jennifer L Roach, J Andrew Whitehead, Gerald J Wyckoff, Kevin J Kolell, Douglas L Crawford
2004 BMC Genomics  
In a case study, a set of genes, that had statistically significant regression between gene expression levels and environmental temperature along the Atlantic Coast, shows a statistically significant (  ...  Primary annotations based on sequence similarity are linked to networks of systematic annotation in Gene Ontology (GO) and the Kyoto Encyclopedia of Genes and Genomes (KEGG) and can be queried and computationally  ...  The application of theses tools in an appropriate framework as outlined in Funny-Base can be used to create a systems level functional genomics annotation system useful for EST databases to study biological  ... 
doi:10.1186/1471-2164-5-96 pmid:15610557 pmcid:PMC544896 fatcat:pilnfl4gyrgahdixsu5wv5s7zy

An experimental evaluation of large scale GBDT systems

Fangeheng Fu, Jiawei Jiang, Yingxia Shao, Bin Cui
2019 Proceedings of the VLDB Endowment  
Based on the analysis, we further propose a novel distributed GBDT system named Vero, which adopts the unexplored composition of vertical partitioning and row-store and suits for many large-scale cases  ...  To validate our analysis empirically, we implement different quadrants in the same code base and compare them under extensive workloads, and finally compare Vero with other state-of-the-art systems over  ...  There is a surge of interests to introduce parameter-server architecture into industrial applications [21, 44, 41] . Notably, TencentBoost and PSMART [20, 43] implement GBDT with parameter-server.  ... 
doi:10.14778/3342263.3342273 fatcat:h3lo7wel25fp3niclkoi2mvrf4

Financial Cybercrime: A Comprehensive Survey of Deep Learning Approaches to Tackle the Evolving Financial Crime Landscape

Jack Nicholls, Aditya Kuppa, Nhien-An Le-Khac
2021 IEEE Access  
They then apply a geographical areas, study the number of transactions, and parameter server based gradient boosted decision tree called finally to identify  ...  Traditionally, rule based systems and shallow anomaly detection methods have been applied to detect financial crime and fraud, but recent developments have seen graph based techniques and neural  ... 
doi:10.1109/access.2021.3134076 fatcat:lm2upcaoabbnbie6r4sfzhjh4y