46,817 Hits in 2.8 sec

Online Robust and Adaptive Learning from Data Streams [article]

Shintaro Fukushima and Atsushi Nitanda and Kenji Yamanishi
2021 arXiv   pre-print
In online learning from non-stationary data streams, it is necessary to learn robustly to outliers and to adapt quickly to changes in the underlying data generating mechanism.  ...  In this paper, we refer to the former attribute of online learning algorithms as robustness and to the latter as adaptivity. There is an obvious tradeoff between the two attributes.  ...  Proposed algorithm In this section, we introduce the proposed online learning algorithm from data streams, called the SRA, to consider the tradeoff between the robustness and adaptivity.  ... 
arXiv:2007.12160v2 fatcat:bv5tolkcafbojcqv75672h6eja

PWPAE: An Ensemble Framework for Concept Drift Adaptation in IoT Data Streams [article]

Li Yang, Dimitrios Michael Manias, Abdallah Shami
2021 arXiv   pre-print
issues often occur in IoT data analytics, as IoT data is often dynamic data streams that change over time, causing model degradation and attack detection failure.  ...  In this paper, we propose a Performance Weighted Probability Averaging Ensemble (PWPAE) framework for drift adaptive IoT anomaly detection through IoT data stream analytics.  ...  To achieve better concept drift adaptation, ensemble learning methods have been proposed to construct robust learners for data stream analytics.  ... 
arXiv:2109.05013v1 fatcat:534y54ne5zhszl3wjwzmllvzhm

A Lightweight Concept Drift Detection and Adaptation Framework for IoT Data Streams [article]

Li Yang, Abdallah Shami
2021 arXiv   pre-print
The proposed adaptive LightGBM model can perform continuous learning and drift adaptation on IoT data streams without human intervention.  ...  A novel drift adaptation method named Optimized Adaptive and Sliding Windowing (OASW) is proposed to adapt to the pattern changes of online IoT data streams.  ...  It comprises two stages: offline learning to obtain an initial trained model, and online training to detect IoT attacks in online data streams.  ... 
arXiv:2104.10529v1 fatcat:tx3or6snhnh6xhbxyab4y2cz6u

Adaptation Strategies for Automated Machine Learning on Evolving Data [article]

Bilge Celik, Joaquin Vanschoren
2021 arXiv   pre-print
These are evaluated empirically on real-world and synthetic data streams with different types of concept drift.  ...  The main goal of this study is to understand the effect of data stream challenges such as concept drift on the performance of AutoML methods, and which adaptation strategies can be employed to make them  ...  ACKNOWLEDGMENTS We would like to thank Erin Ledell, Matthias Feurer and Pieter Gijsbers for their advice on adapting their AutoML systems for this study.  ... 
arXiv:2006.06480v2 fatcat:rw3ggc4ozbfmxeafi26lmzw7ju

Calculating feature importance in data streams with concept drift using Online Random Forest

Andrew Phelps Cassidy, Frank A. Deviney
2014 2014 IEEE International Conference on Big Data (Big Data)  
Numerous researchers have proposed online learning algorithms that train iteratively from new observations, and provide continuously relevant predictions.  ...  Large volume data streams with concept drift have garnered a great deal of attention in the machine learning community.  ...  Initial approaches to learning from data streams with concept drift were based on explicitly detecting, quantifying, and characterizing concept drift and then either re-retraining or re-weighting classifiers  ... 
doi:10.1109/bigdata.2014.7004352 dblp:conf/bigdataconf/CassidyD14 fatcat:luvfadag6jacdnjuos3kjfo4ii

Review paper on adapting data stream mining concept drift using ensemble classifier approach

Nilima Motghare, Arvind Mewada
2014 IOSR Journal of Computer Engineering  
Data stream is massive, fast changing and infinite in nature. It is very natural that large amount of  ...  b) Robustness Issue Noise can severely impair the quality and speed of learning. It is difficult to distinguish noise from changes caused by concept drift.  ...  Adaptive learning algorithms can be seen as advanced incremental learning algorithms that to adapt to evaluation of the data generating process over time.  ... 
doi:10.9790/0661-1654120123 fatcat:wyub5nwbgbfzpgwo5x3hbxwski

Continuous online sequence learning with an unsupervised neural network model [article]

Yuwei Cui, Subutai Ahmad, Jeff Hawkins
2016 arXiv   pre-print
In this paper, we analyze properties of HTM sequence memory and apply it to sequence learning and prediction problems with streaming data.  ...  , robustness to sensor noise and fault tolerance, and good performance without task-specific hyper- parameters tuning.  ...  As a result, the algorithm needs to continuously learn from the data streams and rapidly adapt to changes.  ... 
arXiv:1512.05463v2 fatcat:n2dyllh5bveg3hlvd5slybbgha

Online Semi-supervised Multi-label Classification with Label Compression and Local Smooth Regression

Peiyan Li, Honglian Wang, Christian Böhm, Junming Shao
2020 Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence  
a semi-supervised setting and is robust to evolving label distributions.  ...  Targeting the evolving label distribution problem, we propose an adaptive decoding scheme to adequately integrate newly arriving labeled data.  ...  Fundamental Research Funds for the Central Universities (ZYGX2019Z014), Fok Ying-Tong Education Foundation for Young Teachers in the Higher Education Institutions of China (161062), National key research and  ... 
doi:10.24963/ijcai.2020/189 dblp:conf/ijcai/LiWBS20 fatcat:56nnzcspwrh4xefatldpzhdoti

Adversarial Concept Drift Detection under Poisoning Attacks for Robust Data Stream Mining [article]

Łukasz Korycki, Bartosz Krawczyk
2020 arXiv   pre-print
Continuous learning from streaming data is among the most challenging topics in the contemporary machine learning.  ...  Extensive computational experiments, conducted on both fully and sparsely labeled data streams, prove the high robustness and efficacy of the proposed drift detection framework in adversarial scenarios  ...  Efficient learning from data streams calls for such algorithms that can incorporate new data without a need for being retrained from scratch.  ... 
arXiv:2009.09497v1 fatcat:eybydq3rbbbxjga7nedjglyaze

Special issue on "Data-driven evolutionary optimization"

Yaochu Jin, Jinliang Ding
2017 Soft Computing - A Fusion of Foundations, Methodologies and Applications  
then online ensemble learning adapts to the distribution evolving characteristic of streaming data and overcomes the difficulty of obtaining the optimal hyper-grid structure.  ...  The fourth paper, "Streaming data anomaly detection method based on hyper-grid structure and online ensemble learning" by Ding et al. proposes a novel online streaming data anomaly detection method.  ...  then online ensemble learning adapts to the distribution evolving characteristic of streaming data and overcomes the difficulty of obtaining the optimal hyper-grid structure.  ... 
doi:10.1007/s00500-017-2842-x fatcat:ofzrxnd6cnc2foe2zsc2nlohyi

Introductory Chapter: Data Streams and Online Learning in Social Media [chapter]

Alberto Cano
2020 Social Media and Machine Learning  
Data stream mining for online learning A data stream is an ordered and potentially unbounded sequence of data instances arriving continuously to a machine learning system [13] .  ...  Data stream mining can detect changes in the property of the stream data and adapt the classification model accordingly.  ... 
doi:10.5772/intechopen.90826 fatcat:j3rs4epsbfepbic5iq3twpnpqm

Resampling-Based Ensemble Methods for Online Class Imbalance Learning

Shuo Wang, Leandro L. Minku, Xin Yao
2015 IEEE Transactions on Knowledge and Data Engineering  
Online class imbalance learning is a new learning problem that combines the challenges of both online learning and class imbalance learning.  ...  We find that UOB is better at recognizing minority-class examples in static data streams, and OOB is more robust against dynamic changes in class imbalance status.  ...  Different from incremental learning that processes data in batches, online learning here means learning from data examples "one-by-one" without storing and reprocessing observed examples [3] .  ... 
doi:10.1109/tkde.2014.2345380 fatcat:rgacmhvifbfl7d4wrvwv6qbkry

Online Decision Trees with Fairness [article]

Wenbin Zhang, Liang Zhao
2020 arXiv   pre-print
of all the training data for model learning.  ...  In addition, the data streams might also evolve over time, which further requires the model to be able to simultaneously adapt to non-stationary data distributions and time-evolving bias patterns, with  ...  , thus providing more flexibility than the state of arts. • Two respective fairness-aware learners are designed for learning online from the massive stationary and nonstationary data streams that are common  ... 
arXiv:2010.08146v1 fatcat:tm7rvvrn7fen7koqjbrwuj7ceu

Concept Drift Evolution In Machine Learning Approaches: A Systematic Literature Review

Manzoor Ahmed Hashmani, Syed Muslim Jameel, Mobashar Rehman, Atsushi Inoue
2020 International Journal on Smart Sensing and Intelligent Systems  
However, online machine learning has significant importance to fulfill the demands of the current computing revolution.  ...  Concept Drift's issue is a decisive problem of online machine learning, which causes massive performance degradation in the analysis.  ...  The primary search term Concept Drift, Online Learning and Machine Learning, Adaptive Model, and Big Data are having five derived terms, each shown in Table 2 .  ... 
doi:10.21307/ijssis-2020-029 fatcat:emvpf52xfrbb3okyzj2f4osy2i

Is Fast Adaptation All You Need? [article]

Khurram Javed, Hengshuai Yao, Martha White
2019 arXiv   pre-print
Gradient-based meta-learning has proven to be highly effective at learning model initializations, representations, and update rules that allow fast adaptation from a few samples.  ...  In this paper, we investigate a different training signal -- robustness to catastrophic interference -- and demonstrate that representations learned by directing minimizing interference are more conducive  ...  Standard neural network, without any meta-learning, applied to the CLP task would do poorly as they struggle to learn online from a highly correlated stream of data in a singe pass.  ... 
arXiv:1910.01705v1 fatcat:kv7wbmtcirbvzmbkhov2kvorti
« Previous Showing results 1 — 15 out of 46,817 results