Filters








1,120 Hits in 1.9 sec

Descriptive Dimensionality and Its Characterization of MDL-based Learning and Change Detection [article]

Kenji Yamanishi
2019 arXiv   pre-print
Through the analysis, we demonstrate that Ddim is an intrinsic quantity which characterizes the performance of the MDL-based learning and change detection.  ...  The paper also derives error probabilities of the MDL-based test for multiple model change detection. It proves that they are also governed by Ddim.  ...  We are thus interested in analyzing the relations among the description-based dimensionality and the performance of the MDL-based learning and change detection algorithms.  ... 
arXiv:1910.11540v1 fatcat:sgjan5qmrfaipcbclia5jzvhai

Modeling trajectory of dynamic clusters in image time-series for spatio-temporal reasoning

P. Heas, M. Datcu
2005 IEEE Transactions on Geoscience and Remote Sensing  
The hierarchy is composed of two inference steps: an unsupervised modeling of dynamic clusters resulting in a graph of trajectories, and an interactive learning procedure based on graphs which leads to  ...  This paper presents an information mining concept which enables a user to learn and retrieve spatio-temporal structures in SITS.  ...  Giros for stimulating discussions and for carefully preprocessing the data. The data were made available by the French Space Agency (CNES) from the ADAM project.  ... 
doi:10.1109/tgrs.2005.847791 fatcat:uiarpp3dw5e4figd4fyktlzx7e

Discovering the Intrinsic Cardinality and Dimensionality of Time Series Using MDL

Bing Hu, Thanawin Rakthanmanon, Yuan Hao, Scott Evans, Stefano Lonardi, Eamonn Keogh
2011 2011 IEEE 11th International Conference on Data Mining  
We will frame the discovery of these intrinsic features in the Minimal Description Length (MDL) framework.  ...  In this paper, we investigate techniques to discover the natural intrinsic representation model, dimensionality and alphabet cardinality of a time series.  ...  The first two authors contributed equally and did the bulk of the work, and should be consider joint first authors.  ... 
doi:10.1109/icdm.2011.54 dblp:conf/icdm/HuRHELK11 fatcat:njbuibjnybdi3lm3e43mjlyxfe

Modeling sports highlights using a time-series clustering framework and model interpretation

Regunathan Radhakrishnan, Isao Otsuka, Ziyou Xiong, Ajay Divakaran, Rainer W. Lienhart, Noboru Babaguchi, Edward Y. Chang
2005 Storage and Retrieval Methods and Applications for Multimedia 2005  
The distribution of features from the training data so obtained for each of the key audio classes, is parameterized by a Minimum Description Length Gaussian Mixture Model (MDL-GMM).  ...  In our past work on sports highlights extraction, we have shown the utility of detecting audience reaction using an audio classification framework.  ...  It also implies that it may be sufficient to deal with one background process at a time and detect outliers. . .  ... 
doi:10.1117/12.588059 dblp:conf/spieSR/RadhakrishnanOXD05 fatcat:rz6zfskp5nekfpjni5py2dgm4a

Using the minimum description length to discover the intrinsic cardinality and dimensionality of time series

Bing Hu, Thanawin Rakthanmanon, Yuan Hao, Scott Evans, Stefano Lonardi, Eamonn Keogh
2014 Data mining and knowledge discovery  
In this work, we investigate the problem of discovering the natural intrinsic representation model, dimensionality and alphabet cardinality of a time series.  ...  The ability to automatically discover these intrinsic features has implications beyond selecting the best parameters for particular algorithms, as characterizing data in such a manner is useful in its  ...  Acknowledgments This project was supported by the Department of the United States Air Force, Air Force Research Laboratory under Contract FA8750-10-C-0160, and by NSF grants IIS-1161997.  ... 
doi:10.1007/s10618-014-0345-2 fatcat:jywuttx6s5fhxlu5nlaamczqc4

Representation Edit Distance as a Measure of Novelty [article]

Joshua Alspector
2021 arXiv   pre-print
Adaptation to novelty is viewed as learning to change and augment existing skills to confront unfamiliar situations.  ...  The RED is an intuitive approximation to the change in information content in bit strings measured by comparing pre-novelty and post-novelty skill programs.  ...  Acknowledgments This work was supported by the Science of Artificial Intelligence and Learning for Open-world Novelty (SAIL-ON) program of the U.S. Defense Advanced Research Projects Agency.  ... 
arXiv:2111.02770v1 fatcat:4vdwwy3gdja7xdca7aphnpseta

Unsupervised Discretization by Two-dimensional MDL-based Histogram [article]

Lincen Yang, Mitra Baratchi, Matthijs van Leeuwen
2020 arXiv   pre-print
The state-of-the-art method for one-dimensional data infers locally adaptive histograms using the minimum description length (MDL) principle, but the multi-dimensional case is far less studied: current  ...  We extend the state of the art for the one-dimensional case to obtain a model selection problem based on the normalised maximum likelihood, a form of refined MDL.  ...  two-dimensional histogram as an MDL-based model selection task.  ... 
arXiv:2006.01893v2 fatcat:zi66e6wxtfb4fciyhcn5ajfggu

Concept Drift Detection: Dealing with MissingValues via Fuzzy Distance Estimations [article]

Anjin Liu, Jie Lu, Guangquan Zhang
2020 arXiv   pre-print
We also conjecture that imputation can introduce bias into the process of estimating distribution changes during drift detection, which can make it more difficult to train a learning model.  ...  Experiments on both synthetic and real-world data sets demonstrate the advantages of this method and show its robustness in detecting drift in data with missing values.  ...  Such a change in the drift detection results is dangerous because it may increase the number of false alarms and result in unnecessary warnings.  ... 
arXiv:2008.03662v1 fatcat:q7huujxbsfhpza4vrj6tr3twy4

Natural Language Description of Videos for Smart Surveillance

Aniqa Dilawari, Muhammad Usman Ghani Khan, Yasser D. Al-Otaibi, Zahoor-ur Rehman, Atta-ur Rahman, Yunyoung Nam
2021 Applied Sciences  
Our results show that our framework has distinct advantages over traditional rule-based models for the recognition and generation of natural language descriptions.  ...  This framework is based on the multitask learning of high-level features (HLFs) using a convolutional neural network (CNN) and natural language generation (NLG) through bidirectional recurrent networks  ...  The spatial and temporal features of the video clips were characterized using 2D and 3D dimensional CNN visual features for every video frame in an input clip.  ... 
doi:10.3390/app11093730 fatcat:62o2odoqn5gajk46grynsxaa74

Human activity recognition based on the blob features

Jie Yang, Jian Cheng, Hanqing Lu
2009 2009 IEEE International Conference on Multimedia and Expo  
Then, we use foreground blobs of the current frame and a series of frames before the current frame to form a new feature image in certain rules.  ...  We use Gaussian Mixture to model features for each type of human activities and employ Mahalanobis distance to measure the similarity.  ...  It involves two parameters learning process: background modeling learning and parameters of each mixture model learning in the feature space, and contains three steps to recognize human activity: feature  ... 
doi:10.1109/icme.2009.5202508 dblp:conf/icmcs/YangCL09 fatcat:5fdoz7wtpvdwxobklwtosav2wm

Evaluating Overfit and Underfit in Models of Network Community Structure [article]

Amir Ghasemian, Homa Hosseinmardi, Aaron Clauset
2019 arXiv   pre-print
We introduce a new diagnostic for evaluating overfitting and underfitting in practice, and use it to roughly divide community detection methods into general and specialized learning algorithms.  ...  Across methods and inputs, Bayesian techniques based on the stochastic block model and a minimum description length approach to regularization represent the best general learning approach, but can be outperformed  ...  Across methods, Bayesian and regularized likelihood methods based on SBM tend to perform best, and a minimum description length (MDL) approach to regularization [20] provides the best general learning  ... 
arXiv:1802.10582v3 fatcat:5tk3l27iwvar5hggi4lk6xogpy

The Minimum Description Length Principle for Pattern Mining: A Survey [article]

Esther Galbrun
2021 arXiv   pre-print
After giving an outline of relevant concepts from information theory and coding, as well as of work on the theory behind the MDL and similar principles, we review MDL-based methods for mining various types  ...  This is about the Minimum Description Length (MDL) principle applied to pattern mining. The length of this description is kept to the minimum.  ...  The contents of this survey reflect the understanding of the author, any mistakes and misinterpretations are her own.  ... 
arXiv:2007.14009v3 fatcat:v7zhxwfa5zhc7gd6msyews73ze

Maximum likelihood principle for DNA copy number analysis

Abdullah K. Alqallaf, Ahmed H. Tewfik
2009 2009 IEEE International Conference on Acoustics, Speech and Signal Processing  
Furthermore, we employ the Minimum Description Length rule to estimate the number of unknown parameters.  ...  In this paper, we present a robust procedure for the analysis of DNA copy number data based on maximum likelihood principle using global information of the entire data record.  ...  For this we apply a technique termed the minimum description principle (MDL) presented by [3] (8) where m k is the number of estimated parameters or equivalently the dimensionality of the unknown  ... 
doi:10.1109/icassp.2009.4959630 dblp:conf/icassp/AlqallafT09 fatcat:pgalnq4snnbq7ac527fgsij3wq

Detecting Change Signs with Differential MDL Change Statistics for COVID-19 Pandemic Analysis [article]

Kenji Yamanishi, Linchuan Xu, Ryo Yuki, Shintaro Fukushima, Chuan-hao Lin
2021 arXiv   pre-print
We are concerned with the issue of detecting changes and their signs from a data stream.  ...  The key idea is to employ a new information-theoretic notion, which we call the differential minimum description length change statistics (D-MDL), for measuring the scores of change sign.  ...  We propose algorithms for on-line change sign detection based on D-MDL. (2) Theoretical and empirical justification of D-MDL.  ... 
arXiv:2007.15179v2 fatcat:wqnizowxzrhbbogydpxzddl3s4

Unsupervised Idealization of Nano-Electronic Sensors Recordings with Concept Drifts: A Compressive Feature Learning Approach for Non-Stationary Single-Molecule Data Analysis [article]

Mohamed Ouqamra
2020 bioRxiv   pre-print
We present a new smFET data analysis framework, based on a compressive feature learning scheme to optimize unsupervised idealization of smFET traces, by a precise and accurate molecular events detection  ...  Single-molecule nanocircuits based on field-effect transistors (smFETs) are emerging and promising nano-bioelectronic sensors for the functional detection of molecular dynamics involved in biochemical  ...  Another approach based on the minimum description length (MDL) principle has been proposed for the idealization of ion channel recordings (40) .  ... 
doi:10.1101/2020.05.02.074013 fatcat:uybuvdynajaaxjl46tr23r7xye
« Previous Showing results 1 — 15 out of 1,120 results