Within the domain of patternrecognition, the automated identification of handwritten characters or symbols presents a complex handwriting recognition challenges. In this paper, a novel methodology is presented, which...
详细信息
The rise of Artificial Intelligence (AI) and robotics in the past decade has created various career opportunities in many industries such as robotics, manufacturing and healthcare. Thus, skills such as AI and robotics...
详细信息
Facial expression recognition aims to classify human expression using the designed algorithm. Recently, the popular FER2013dataset suffered several bottlenecks for learning a good model. In this paper, we aim to desi...
详细信息
Many applications require predictions with confidence. We are interested in Confidence machines which are algorithms that call provide some measure oil how confident they are that their Output is correct. Confidence M...
详细信息
ISBN:
(纸本)9783642030697
Many applications require predictions with confidence. We are interested in Confidence machines which are algorithms that call provide some measure oil how confident they are that their Output is correct. Confidence machines are quite general and there are many algorithms solving the problem of prediction with confidence. As predictors we consider Venn Probability machines and Conformal Predictors. Both of these algorithms rely oil all underlying algorithm for prediction and in this paper We use two simple algorithms, namely the Nearest Neighbours and Nearest Centroid algorithms. Our aim is to provide some guidelines on how to choose the Most Suitable algorithm for a practical application where confidence is needed.
This paper shows a preliminary report regarding classification techniques based on argumentation theory in artificial intelligence. A classification problem is defined on a directed graph, i.e., an argumentation frame...
详细信息
ISBN:
(纸本)9781479942749
This paper shows a preliminary report regarding classification techniques based on argumentation theory in artificial intelligence. A classification problem is defined on a directed graph, i.e., an argumentation framework, where each node represents an argument and each edge an attack relation between connected arguments. A hypothesis space is defined by all possible argumentation consequences, i.e., extensions. A target argument is classified as justified or overruled, according to the best extensions minimizing errors with respect to training examples, i.e.,tuples of arguments and their correct classes. We give ideal downward and upward refinement operators for calculating hypotheses step by step. Algorithm analysis and performance evaluation are future work.
Actitracker is a smartphone-based activity-monitoring service to help people ensure they receive sufficient activity to maintain proper health. This free service allowed people to set personal activity goals and monit...
详细信息
ISBN:
(纸本)9781509052066
Actitracker is a smartphone-based activity-monitoring service to help people ensure they receive sufficient activity to maintain proper health. This free service allowed people to set personal activity goals and monitor their progress toward these goals. Actitracker uses machinelearning methods to recognize a user's activities. It initially employs a "universal" model generated from labeled activity data from a panel of users, but will automatically shift to a much more accurate personalized model once a user completes a simple training phase. Detailed activity reports and statistics are maintained and provided to the user. Actitracker is a research-based system that began in 2011, before fitness trackers like Fitbit were popular, and was deployed for public use from 2012 until 2015, during which period it had 1,000 registered users. This paper describes the Actitracker system, its use of machinelearning, and user experiences. While activity recognition has now entered the mainstream, this paper provides insights into applied activity recognition, something that commercial companies rarely share.
The field of medical analysis is often referred to be a valuable source of rich information. Coronary Heart Disease (CHD) is one of the major causes of death all around the world therefore early detection of CHD can h...
详细信息
ISBN:
(纸本)9781450371605
The field of medical analysis is often referred to be a valuable source of rich information. Coronary Heart Disease (CHD) is one of the major causes of death all around the world therefore early detection of CHD can help reduce these rates. The challenge lies in the complexity of the data and correlations when it comes to prediction using conventional techniques. The aim of this research is to use the historical medical data to predict CHD using machinelearning (ML) technology. The scope of this research is limited to using three supervised learning techniques namely Naive Bayes (NB), Support Vector machine (SVM) and Decision Tree (DT), to discover correlations in CHD data that might help improving the prediction rate. Using the South African Heart Disease dataset of 462 instances, intelligent models are derived by the considered ML techniques using 10-fold cross validation. Empirical results using different performance evaluation measures report that probabilistic models derived by NB are promising in detecting CHD.
In this research, we developed a new deep neural network model to identify human action that was composed of an autoencoder and a patternrecognition neural network (PRNN). Our approach was divided into two parts: a s...
详细信息
ISBN:
(纸本)9781509063529
In this research, we developed a new deep neural network model to identify human action that was composed of an autoencoder and a patternrecognition neural network (PRNN). Our approach was divided into two parts: a system learning stage and an action recognition stage. In the system learning stage, first we secured human body outlines for each image frame, and combined the outlines to build an overlay of binary images to use as training data. Based on deep neural network learning, an autoencoder was trained to extract action features. Next, we used supervised learning to train a PRNN on the obtained features. Last, we combined the autoencoder with the PRNN to build a new deep neural network called the APRNN. Using fine tuning, the APRNN achieved optimal performance. In the action recognition stage of our approach, human action sequences were translated into binary overlay images, and the ARPNN was used to identify the actions. Test results showed our method had better performance than existing approaches.
Clustering technique is an important tool for data analysis and has a promising prospect in datamining, patternrecognition, etc. Usually, objects in clustering analysis are of vectors, which consist of some features...
详细信息
ISBN:
(纸本)0769528759
Clustering technique is an important tool for data analysis and has a promising prospect in datamining, patternrecognition, etc. Usually, objects in clustering analysis are of vectors, which consist of some features.. They may be represented as points in Euclidean space. However, in some tasks, objects in clustering analysis may be some abstract models other than data points, for example neural networks, decision trees, support vector machines, etc. By defining the extended distance (in real tasks, there are some different definition forms about distance), clustering method is studied for the abstract data objects. Framework of clustering algorithm for objects of models is presented As its application, a method for improving diversity of ensemble learning with neural networks is investigated. The relations between the number of clusters in clustering analysis, the size of ensemble learning, and performance of ensemble learning are studied by experiments.
The CS framework is useful for a much wide range of patternrecognition tasks such as visual object classification. It is possible to directly extract features from a small number of random projections without ever re...
详细信息
ISBN:
(纸本)9781509025350
The CS framework is useful for a much wide range of patternrecognition tasks such as visual object classification. It is possible to directly extract features from a small number of random projections without ever reconstructing the signal, which results in compressed learning. As for compressed learning, it is to learn with randomly projected data, compressed data, instead of original data. learning with compressed data saves considerable running time and storage since random projection can effectively reduce the dimension of data. In this paper, works dealing with compressed data concentrate on the adaboost classification case. It has been verified by the experiments that the possibility of AdaBoost algorithm learning in compressed space, and a better test error result has been acquired although only the simple stump is adopted as the weak classifiers.
暂无评论