Anzah H. Niazi, D. Yazdansepas, Jennifer L. Gay, Frederick W. Maier, Lakshmish Ramaswamy, K. Rasheed, M. Buman
This paper proposes a multi-level meta-classifier for identifying human activities based on accelerometer data. The training data consists of 77 subjects performing a combination of 23 different activities and monitored using a single hip-worn triaxial accelerometer. Time and frequency based features were extracted from two-second windows of raw accelerometer data and a subset of the features, together with demographic information, was selected for classification. The activities were divided into five activity groups: non-ambulatory activities, walking, running, climbing upstairs, and climbing downstairs. Multiple classification techniques were tested for each classifier level and groups. Random forests were found to perform comparatively better at each level. Based upon those tests, a 3-level hierarchical classifier, consisting of 5 random forest classifiers, was built. At the first level, the non-ambulatory activities are separated from the rest. At the second, the ambulatory activities are divided into four activity groups. At the final level, the activities are classified individually. Accuracy on test sets was found to be approximately 87% overall for individual activities and 94% at the activity group level. These results compare favorably to contemporary results in classifying human activity.
{"title":"A Hierarchical Meta-Classifier for Human Activity Recognition","authors":"Anzah H. Niazi, D. Yazdansepas, Jennifer L. Gay, Frederick W. Maier, Lakshmish Ramaswamy, K. Rasheed, M. Buman","doi":"10.1109/ICMLA.2016.0022","DOIUrl":"https://doi.org/10.1109/ICMLA.2016.0022","url":null,"abstract":"This paper proposes a multi-level meta-classifier for identifying human activities based on accelerometer data. The training data consists of 77 subjects performing a combination of 23 different activities and monitored using a single hip-worn triaxial accelerometer. Time and frequency based features were extracted from two-second windows of raw accelerometer data and a subset of the features, together with demographic information, was selected for classification. The activities were divided into five activity groups: non-ambulatory activities, walking, running, climbing upstairs, and climbing downstairs. Multiple classification techniques were tested for each classifier level and groups. Random forests were found to perform comparatively better at each level. Based upon those tests, a 3-level hierarchical classifier, consisting of 5 random forest classifiers, was built. At the first level, the non-ambulatory activities are separated from the rest. At the second, the ambulatory activities are divided into four activity groups. At the final level, the activities are classified individually. Accuracy on test sets was found to be approximately 87% overall for individual activities and 94% at the activity group level. These results compare favorably to contemporary results in classifying human activity.","PeriodicalId":356182,"journal":{"name":"2016 15th IEEE International Conference on Machine Learning and Applications (ICMLA)","volume":"64 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123952634","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In this paper, the different machine learning and data mining approaches used for Residential Energy Smart Management (RESM) will be discussed and classified according to some meaningful criteria. The proposed classification is an attempt to highlight the advantages and limitations of each category. Moreover, we emphasize the complementarity between approaches belonging to different categories and we point out the main challenges that still face RESM.
{"title":"A Review on Machine Learning and Data Mining Techniques for Residential Energy Smart Management","authors":"Hajer Salem, M. S. Mouchaweh, A. Hassine","doi":"10.1109/ICMLA.2016.0195","DOIUrl":"https://doi.org/10.1109/ICMLA.2016.0195","url":null,"abstract":"In this paper, the different machine learning and data mining approaches used for Residential Energy Smart Management (RESM) will be discussed and classified according to some meaningful criteria. The proposed classification is an attempt to highlight the advantages and limitations of each category. Moreover, we emphasize the complementarity between approaches belonging to different categories and we point out the main challenges that still face RESM.","PeriodicalId":356182,"journal":{"name":"2016 15th IEEE International Conference on Machine Learning and Applications (ICMLA)","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123994890","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Kernel density estimation is a popular method for identifying crime hotspots for the purpose of data-driven policing. However, computing a kernel density estimate is computationally intensive for large crime datasets, and the quality of the resulting estimate depends heavily on parameters that are difficult to set manually. Inspired by methods from image processing, we propose a novel way for performing hotspot analysis using localized kernel density estimation optimized with an evolutionary algorithm. The proposed method uses local learning to address three challenges associated with traditional kernel density estimation: computational complexity, bandwidth selection, and kernel function selection. We evaluate our localized kernel model on 17 crime types from Chicago, Illinois, USA. Preliminary results indicate significant improvement in prediction performance over the traditional approach. We also examine the effect of data sparseness on the performance of both models.
{"title":"Automatic Optimization of Localized Kernel Density Estimation for Hotspot Policing","authors":"Mohammad Al Boni, M. Gerber","doi":"10.1109/ICMLA.2016.0015","DOIUrl":"https://doi.org/10.1109/ICMLA.2016.0015","url":null,"abstract":"Kernel density estimation is a popular method for identifying crime hotspots for the purpose of data-driven policing. However, computing a kernel density estimate is computationally intensive for large crime datasets, and the quality of the resulting estimate depends heavily on parameters that are difficult to set manually. Inspired by methods from image processing, we propose a novel way for performing hotspot analysis using localized kernel density estimation optimized with an evolutionary algorithm. The proposed method uses local learning to address three challenges associated with traditional kernel density estimation: computational complexity, bandwidth selection, and kernel function selection. We evaluate our localized kernel model on 17 crime types from Chicago, Illinois, USA. Preliminary results indicate significant improvement in prediction performance over the traditional approach. We also examine the effect of data sparseness on the performance of both models.","PeriodicalId":356182,"journal":{"name":"2016 15th IEEE International Conference on Machine Learning and Applications (ICMLA)","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123114082","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This works deals with the concept of liver segmentation by using a priori information based on probabilistic atlases and segmentation learning based of previous steps. A probabilistic atlas is here understood as a probability or membership map that tells how likely is that a point belongs to a shape drawn from the shape distribution at hand. We devise a procedure to segment Perfusion Magnetic Resonance liver images that combines both: a probabilistic atlas of the liver and a segmentation algorithm based on global information of previous simpler segmentation steps, local information from close segmented slices and finally a mathematical morphology procedure, namely viscous reconstruction, to fill the shape. Preliminary results of the algorithm are provided.
{"title":"Iteratively Learning a Liver Segmentation Using Probabilistic Atlases: Preliminary Results","authors":"J. Domingo, E. Durá, Evgin Göçeri","doi":"10.1109/ICMLA.2016.0104","DOIUrl":"https://doi.org/10.1109/ICMLA.2016.0104","url":null,"abstract":"This works deals with the concept of liver segmentation by using a priori information based on probabilistic atlases and segmentation learning based of previous steps. A probabilistic atlas is here understood as a probability or membership map that tells how likely is that a point belongs to a shape drawn from the shape distribution at hand. We devise a procedure to segment Perfusion Magnetic Resonance liver images that combines both: a probabilistic atlas of the liver and a segmentation algorithm based on global information of previous simpler segmentation steps, local information from close segmented slices and finally a mathematical morphology procedure, namely viscous reconstruction, to fill the shape. Preliminary results of the algorithm are provided.","PeriodicalId":356182,"journal":{"name":"2016 15th IEEE International Conference on Machine Learning and Applications (ICMLA)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130766521","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Multi-Attribute Reverse Auctions (MARAs) are considered an excellent way to buy and sell efficiently. However, eliciting the buyer's requirements and preferences as well as determining the winner, are both challenging tasks. In this paper, we propose a multi-round and semi-sealed MARA auction system, capable of determining the winner given a set of user's preferences and requirements. This system is capable of managing qualitative, quantitative and conditional preferences together with constraints. For that, we use the constrained Tradeoffs-enhanced Conditional Preference Networks (constrained TCP-nets) graphical model for representing constraints as well as qualitative and conditional preferences, and Multi-Attribute Utility Theory (MAUT) for dealing with quantitative preferences. Determining the winners of the auction will then be achieved using the backtrack search algorithm we use for solving constrained TCP-nets.
{"title":"Managing Constraints and Preferences for Winner Determination in Multi-attribute Reverse Auctions","authors":"Malek Mouhoub, Farnaz Ghavamifar","doi":"10.1109/ICMLA.2016.0181","DOIUrl":"https://doi.org/10.1109/ICMLA.2016.0181","url":null,"abstract":"Multi-Attribute Reverse Auctions (MARAs) are considered an excellent way to buy and sell efficiently. However, eliciting the buyer's requirements and preferences as well as determining the winner, are both challenging tasks. In this paper, we propose a multi-round and semi-sealed MARA auction system, capable of determining the winner given a set of user's preferences and requirements. This system is capable of managing qualitative, quantitative and conditional preferences together with constraints. For that, we use the constrained Tradeoffs-enhanced Conditional Preference Networks (constrained TCP-nets) graphical model for representing constraints as well as qualitative and conditional preferences, and Multi-Attribute Utility Theory (MAUT) for dealing with quantitative preferences. Determining the winners of the auction will then be achieved using the backtrack search algorithm we use for solving constrained TCP-nets.","PeriodicalId":356182,"journal":{"name":"2016 15th IEEE International Conference on Machine Learning and Applications (ICMLA)","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127020300","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Brain lesions are life threatening diseases. Traditional diagnosis of brain lesions is performed visually by neuro-radiologists. Nowadays, advanced technologies and the progress in magnetic resonance imaging provide computer aided diagnosis using automated methods that can detect and segment abnormal regions from different medical images. Among several techniques, machine learning based methods are flexible and efficient. Therefore, in this paper, we present a review on techniques applied for detection and segmentation of brain lesions from magnetic resonance images with supervised and unsupervised machine learning techniques.
{"title":"Review on Machine Learning Based Lesion Segmentation Methods from Brain MR Images","authors":"Evgin Göçeri, E. Durá, M. Günay","doi":"10.1109/ICMLA.2016.0102","DOIUrl":"https://doi.org/10.1109/ICMLA.2016.0102","url":null,"abstract":"Brain lesions are life threatening diseases. Traditional diagnosis of brain lesions is performed visually by neuro-radiologists. Nowadays, advanced technologies and the progress in magnetic resonance imaging provide computer aided diagnosis using automated methods that can detect and segment abnormal regions from different medical images. Among several techniques, machine learning based methods are flexible and efficient. Therefore, in this paper, we present a review on techniques applied for detection and segmentation of brain lesions from magnetic resonance images with supervised and unsupervised machine learning techniques.","PeriodicalId":356182,"journal":{"name":"2016 15th IEEE International Conference on Machine Learning and Applications (ICMLA)","volume":"73 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126515604","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In this research, we have developed a model for predicting the profitability class of a movie namely "Profit" and "Loss" based on the data about movies released between the years 2010 and 2015. Our methodology considers both historical data as well as data extracted from the social media. This data is normalized and then given a weight using standard normalization techniques. The cleaned and normalized dataset is then used to train a back-propagation cross entropy validated neural network. Results show that our strategy of identifying the class of success is highly effective and accurate when compared to the results from using a support machine vector on the data.
{"title":"Predicting Movie Box Office Profitability: A Neural Network Approach","authors":"Travis Ginmu Rhee, F. Zulkernine","doi":"10.1109/ICMLA.2016.0117","DOIUrl":"https://doi.org/10.1109/ICMLA.2016.0117","url":null,"abstract":"In this research, we have developed a model for predicting the profitability class of a movie namely \"Profit\" and \"Loss\" based on the data about movies released between the years 2010 and 2015. Our methodology considers both historical data as well as data extracted from the social media. This data is normalized and then given a weight using standard normalization techniques. The cleaned and normalized dataset is then used to train a back-propagation cross entropy validated neural network. Results show that our strategy of identifying the class of success is highly effective and accurate when compared to the results from using a support machine vector on the data.","PeriodicalId":356182,"journal":{"name":"2016 15th IEEE International Conference on Machine Learning and Applications (ICMLA)","volume":"261 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114468860","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Different types of Bayesian networks may be used for supervised classification. We combine such approaches together with feature selection and discretization and we show that such combination gives rise to powerful classifiers. A large choice of data sets from the UCI machine learning repository are used in our experiments and an application to Epilepsy type prediction based on PET scan data confirms the efficiency of our approach.
{"title":"Bayesian Network Classification: Application to Epilepsy Type Prediction Using PET Scan Data","authors":"Kamel Jebreen, B. Ghattas","doi":"10.1109/ICMLA.2016.0174","DOIUrl":"https://doi.org/10.1109/ICMLA.2016.0174","url":null,"abstract":"Different types of Bayesian networks may be used for supervised classification. We combine such approaches together with feature selection and discretization and we show that such combination gives rise to powerful classifiers. A large choice of data sets from the UCI machine learning repository are used in our experiments and an application to Epilepsy type prediction based on PET scan data confirms the efficiency of our approach.","PeriodicalId":356182,"journal":{"name":"2016 15th IEEE International Conference on Machine Learning and Applications (ICMLA)","volume":"63 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115913191","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Patrick M. Monamo, Vukosi Marivate, Bhesipho Twala
In the Bitcoin network, lack of class labels tend to cause obscurities in anomalous financial behaviour interpretation. To understand fraud in the latest development of the financial sector, a multifaceted approach is proposed. In this paper, Bitcoin fraud is described from both global and local perspectives using trimmed k-means and kd-trees. The two spheres are investigated further through random forests, maximum likelihood-based and boosted binary regression models. Although both angles show good performance, global outlier perspective outperforms the local viewpoint with exception of random forest that exhibits nearby perfect results from both dimensions. This signifies that features extracted for this study describe the network fairly.
{"title":"A Multifaceted Approach to Bitcoin Fraud Detection: Global and Local Outliers","authors":"Patrick M. Monamo, Vukosi Marivate, Bhesipho Twala","doi":"10.1109/ICMLA.2016.0039","DOIUrl":"https://doi.org/10.1109/ICMLA.2016.0039","url":null,"abstract":"In the Bitcoin network, lack of class labels tend to cause obscurities in anomalous financial behaviour interpretation. To understand fraud in the latest development of the financial sector, a multifaceted approach is proposed. In this paper, Bitcoin fraud is described from both global and local perspectives using trimmed k-means and kd-trees. The two spheres are investigated further through random forests, maximum likelihood-based and boosted binary regression models. Although both angles show good performance, global outlier perspective outperforms the local viewpoint with exception of random forest that exhibits nearby perfect results from both dimensions. This signifies that features extracted for this study describe the network fairly.","PeriodicalId":356182,"journal":{"name":"2016 15th IEEE International Conference on Machine Learning and Applications (ICMLA)","volume":"80 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116939271","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Yoshitomo Matsubara, H. Nishimura, T. Samura, Hiroyuki Yoshimoto, Ryohei Tanimoto
Physical biometrics technologies are introduced to the login process on smart devices. However, many of them have several disadvantages: requirement of embedding special sensor, limited environment to use and copy of key information for authentication. In this research, we proposed a new biometrics technique which can capture user's inimitable behavioral features in his/her spontaneous flick reactions on a touch-screen display for unlocking the device when it wakes up. For practical use of the technique, we adopted one-class classification approaches and they achieved about 1-2% EERs for 2500 samples from 50 subjects.
{"title":"Screen Unlocking by Spontaneous Flick Reactions with One-Class Classification Approaches","authors":"Yoshitomo Matsubara, H. Nishimura, T. Samura, Hiroyuki Yoshimoto, Ryohei Tanimoto","doi":"10.1109/ICMLA.2016.0134","DOIUrl":"https://doi.org/10.1109/ICMLA.2016.0134","url":null,"abstract":"Physical biometrics technologies are introduced to the login process on smart devices. However, many of them have several disadvantages: requirement of embedding special sensor, limited environment to use and copy of key information for authentication. In this research, we proposed a new biometrics technique which can capture user's inimitable behavioral features in his/her spontaneous flick reactions on a touch-screen display for unlocking the device when it wakes up. For practical use of the technique, we adopted one-class classification approaches and they achieved about 1-2% EERs for 2500 samples from 50 subjects.","PeriodicalId":356182,"journal":{"name":"2016 15th IEEE International Conference on Machine Learning and Applications (ICMLA)","volume":"409 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2016-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123537336","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：