Filters








138 Hits in 5.9 sec

On Ensemble SSL Algorithms for Credit Scoring Problem

Ioannis Livieris, Niki Kiriakidou, Andreas Kanavos, Vassilis Tampakas, Panagiotis Pintelas
2018 Informatics  
In this work, we evaluate the performance of two ensemble semi-supervised learning algorithms for the credit scoring problem.  ...  Credit scoring is generally recognized as one of the most significant operational research techniques used in banking and finance, aiming to identify whether a credit consumer belongs to either a legitimate  ...  Conclusions In this work, we evaluated the performance of two ensemble SSL algorithms, entitled CST-Voting and EnSSL, for the credit scoring problem.  ... 
doi:10.3390/informatics5040040 fatcat:h2z2mhie5barlm4mf2lnoirrai

XGBoost Optimized by Adaptive Particle Swarm Optimization for Credit Scoring

Chao Qin, Yunfeng Zhang, Fangxun Bao, Caiming Zhang, Peide Liu, Peipei Liu, Sotiris B. Kotsiantis
2021 Mathematical Problems in Engineering  
To solve this problem, this paper proposes an eXtreme Gradient Boosting credit scoring model that is based on adaptive particle swarm optimization.  ...  Personal credit scoring is a challenging issue. In recent years, research has shown that machine learning has satisfactory performance in credit scoring.  ...  bankruptcy data, which also shows that SSL is feasible for solving the same problem of financial credit scoring.  ... 
doi:10.1155/2021/6655510 fatcat:64wvmu7kgjgv3k3hrrq2a5pfe4

Robust Deep Semi-Supervised Learning: A Brief Introduction [article]

Lan-Zhe Guo and Zhi Zhou and Yu-Feng Li
2022 arXiv   pre-print
Finally, we propose possible promising directions within robust SSL to provide insights for future research.  ...  Recently, SSL with deep models has proven to be successful on standard benchmark tasks.  ...  [67] proposed an open-set example detection score based on the ensemble of model predictions.  ... 
arXiv:2202.05975v1 fatcat:g6ra3ciljbhera44w5bxqqqvhu

DeepFIB: Self-Imputation for Time Series Anomaly Detection [article]

Minhao Liu, Zhijian Xu, Qiang Xu
2021 arXiv   pre-print
For continuous outliers, we also propose an anomaly localization algorithm that dramatically reduces AD errors.  ...  Experiments on various real-world TS datasets demonstrate that DeepFIB outperforms state-of-the-art methods by a large margin, achieving up to 65.2% relative improvement in F1-score.  ...  DeepFIB is a simple yet effective SSL technique to tackle the above problem.  ... 
arXiv:2112.06247v1 fatcat:k7gmfcsnajafdp5qgwq4iuq4f4

A Survey on Semi-Supervised Learning for Delayed Partially Labelled Data Streams [article]

Heitor Murilo Gomes, Maciej Grzenda, Rodrigo Mello, Jesse Read, Minh Huong Le Nguyen, Albert Bifet
2021 arXiv   pre-print
To address the learning problems associated with such data, one can ignore the unlabelled data and focus only on the labelled data (supervised learning); use the labelled data and attempt to leverage the  ...  We propose a unified problem setting, discuss the learning guarantees and existing methods, explain the differences between related problem settings.  ...  One possible pathway for future research is the application of such SSL techniques in regression problems. Fig. 1 . 1 Learning from data streams according to labels arrival time, based on [41] .  ... 
arXiv:2106.09170v1 fatcat:nn4ja4ptsndwngyocjne2tp4ae

Confident Sinkhorn Allocation for Pseudo-Labeling [article]

Vu Nguyen and Sachin Farfade and Anton van den Hengel
2022 arXiv   pre-print
This paper addresses this problem by proposing a Confident Sinkhorn Allocation (CSA), which assigns labels to only samples with high confidence scores and learns the best label allocation via optimal transport  ...  Semi-supervised learning is a critical tool in reducing machine learning's dependence on labeled data.  ...  To highlight the role of variances and data noises toward SSL, we consider a simple binary classification problem on one-dimensional space.  ... 
arXiv:2206.05880v1 fatcat:2bh2j3zwkjfr3ew3edlv7ry6ne

Fair-SSL: Building fair ML Software with less data [article]

Joymallya Chakraborty, Suvodeep Majumder, Huy Tu
2022 arXiv   pre-print
After experimenting on ten datasets and three learners, we find that Fair-SSL achieves similar performance as three state-of-the-art bias mitigation algorithms.  ...  Our framework, Fair-SSL, takes a very small amount (10%) of labeled data as input and generates pseudo-labels for the unlabeled data.  ...  They used self-training (a type of SSL) with ensemble models to generate fair results.  ... 
arXiv:2111.02038v4 fatcat:4r5b45bxeval3haexdop7fiypu

Addressing cold start in recommender systems

Mi Zhang, Jie Tang, Xuchen Zhang, Xiangyang Xue
2014 Proceedings of the 37th international ACM SIGIR conference on Research & development in information retrieval - SIGIR '14  
The proposed algorithms are evaluated on two real-world datasets.  ...  Cold start is one of the most challenging problems in recommender systems. In this paper we tackle the cold-start problem by proposing a context-aware semi-supervised co-training method named C-SEL.  ...  Figure 3 : 3 Comparison between the FactCF algorithm (left) and CSELv (right) on the cold-start problem for D ′ 2 . The average RMSE scores on user bins with different popularity are given.  ... 
doi:10.1145/2600428.2609599 dblp:conf/sigir/ZhangTZX14 fatcat:mle3bv4gmfgobm65jyjxzpgqle

Investigation of Combining Logitboost(M5P) under Active Learning Classification Tasks

Vangjel Kazllarof, Stamatis Karlos, Sotiris Kotsiantis
2020 Informatics  
For this reason, we investigate the use of the Logitboost wrapper classifier, a popular variant of ensemble algorithms which adopts the technique of boosting along with a regression base learner based  ...  efficacy of one hyperparameter of the proposed algorithm.  ...  : the proposed algorithm versus the ensemble ones.  ... 
doi:10.3390/informatics7040050 fatcat:ta5zgkf25vetrmomp5vh5jgkla

Decision making via semi-supervised machine learning techniques [article]

Eftychios Protopapadakis
2016 arXiv   pre-print
Semi-supervised learning (SSL) is a class of supervised learning tasks and techniques that also exploits the unlabeled data for training.  ...  Decision support systems (DSSs) who utilize SSL have significant advantages. Only a small amount of labelled data is required for the initialization.  ...  (Abdou and Pointon, 2011) review 214 previous studies on credit scoring applications, emphasizing on the statistical techniques used for evaluation.  ... 
arXiv:1606.09022v1 fatcat:zzghg6tsijgszaduo7il2rixly

Financial fraud detection by using Grammar-based multi-objective genetic programming with ensemble learning

Haibing Li, Man-Leung Wong
2015 2015 IEEE Congress on Evolutionary Computation (CEC)  
In general, traditional modeling approaches are applied and based on pre-defined hypothesis testing of causes and effects for FFD problems.  ...  Third, it provides an efficient multi-objective method for solving FFD problems.  ...  Therefore, we propose an ensemble method called statistical selection learning (SSL). The idea of the SSL is straightforward.  ... 
doi:10.1109/cec.2015.7257014 dblp:conf/cec/LiW15 fatcat:bn72akt7cbbu7dkpssyee63ac4

Semi-supervised Learning with Deep Generative Models for Asset Failure Prediction [article]

Andre S. Yoon, Taehoon Lee, Yongsub Lim, Deokwoo Jung, Philgyun Kang, Dongwon Kim, Keuntae Park, Yongjin Choi
2017 arXiv   pre-print
The proposed method is evaluated on a publicly available dataset for remaining useful life (RUL) estimation, which shows significant improvement even when a fraction of the data with known health status  ...  It allows us to build prognostic models with the limited amount of health status information for the precise prediction of future asset reliability.  ...  In our study, we use a single iteration since we do not impose any convergence condition for the algorithm. Algorithm 1: Semi-supervised learning based on nonlinear embedding with VAE.  ... 
arXiv:1709.00845v1 fatcat:3jdrhdpoeje23ftwwy7wodhemu

LaserMix for Semi-Supervised LiDAR Semantic Segmentation [article]

Lingdong Kong and Jiawei Ren and Liang Pan and Ziwei Liu
2022 arXiv   pre-print
In this work, we study the underexplored semi-supervised learning (SSL) in LiDAR segmentation.  ...  Notably, we achieve competitive results over fully-supervised counterparts with 2x to 5x fewer labels and improve the supervised-only baseline significantly by 10.8% on average.  ...  for different SSL algorithms on the val set of nuScenes[15].  ... 
arXiv:2207.00026v1 fatcat:clms7vezpfgutgokgxvj5tubhe

Learning from evolving video streams in a multi-camera scenario

Samaneh Khoshrou, Jaime S. Cardoso, Luís F. Teixeira
2015 Machine Learning  
Acknowledgments The authors would like to thank Fundação para a Ciência e a Tecnologia (FCT)-Portugal for financing this work through the grant SFRH/BD/80013/2011.  ...  The algorithm is an one-pass class-based ensemble of classifiers that trains a separate model (h j t ) for a class j at every time slot t.  ...  In order to get a decision on a frame (assign a score to the frame), the outputs of the models are combined using a weighted strategy that gives more credit to the more recent knowledge.  ... 
doi:10.1007/s10994-015-5515-y fatcat:ivg2jg7eszcptmnfhs6hdefphy

A machine learning solution to assess privacy policy completeness

Elisa Costante, Yuanhao Sun, Milan Petković, Jerry den Hartog
2012 Proceedings of the 2012 ACM workshop on Privacy in the electronic society - WPES '12  
Since the algorithm selection problem, i.e. the problem of selecting the algorithm that performs best for a given task, is still unsolved [29] , we test the performance of different algorithms to determine  ...  The CSVi score for classifying documents may represent e.g. a probability, or a measure of vector closeness [40] depending on the algorithm used.  ... 
doi:10.1145/2381966.2381979 dblp:conf/wpes/CostanteSPH12 fatcat:dndzxdvpezh3vdbxarh24eclzy
« Previous Showing results 1 — 15 out of 138 results