If we assume that a natural image can be modeled as a succession of a multilevel system, we can develop an optimal routine of a matrix superposition. Each matrix separates the fundamental elements by a set of optimal ...
详细信息
ISBN:
(纸本)9781628410273
If we assume that a natural image can be modeled as a succession of a multilevel system, we can develop an optimal routine of a matrix superposition. Each matrix separates the fundamental elements by a set of optimal criteria. The matrix superposition is then characterized by a tree-based principle which is applied adaptively. We will also demonstrate how the missing data constrains may be overcome by collecting additional measurements.
Survival analysis models time-to-event distributions with censorship. Recently, deep survival models using neural networks have dominated due to their representational power and state-of-the-art performance. However, ...
详细信息
ISBN:
(纸本)9798400704369
Survival analysis models time-to-event distributions with censorship. Recently, deep survival models using neural networks have dominated due to their representational power and state-of-the-art performance. However, their "black-box" nature hinders interpretability, which is crucial in real-world applications. In contrast, "white-box" tree-based survival models offer better interpretability but struggle to converge to global optima due to greedy expansion. In this paper, we bridge the gap between previous deep survival models and traditional tree-based survival models through deep rectified linear unit (ReLU) networks. We show that a deliberately constructed deep ReLU network (termed SurvReLU) can harness the interpretability of tree-based structures with the representational power of deep survival models. Empirical studies on both simulated and real survival benchmark datasets showed the effectiveness of the proposed SurvReLU in terms of performance and interoperability. The code is available at https://***/xs018/SurvReLU.
The present approach combines data fusion from several sensor types to enhance the overall detection and classification performance. The fusion of different sensors is implemented at data and feature levels that resul...
详细信息
ISBN:
(纸本)9781628415926
The present approach combines data fusion from several sensor types to enhance the overall detection and classification performance. The fusion of different sensors is implemented at data and feature levels that results in enhanced target identification by the means of spatial spectral analysis.
Health disparity is an important public health policy concern as it is related to the social inequalities in population health, not only for the persons experiencing them, but also for the entire population in the soc...
详细信息
Health disparity is an important public health policy concern as it is related to the social inequalities in population health, not only for the persons experiencing them, but also for the entire population in the society. People experiencing poorer health status will negatively impact the overall health of the nation and this inequality is costly as well as burdensome to healthcare system. Health disparities studies allow us to identify and understand disparities, eventually design intervention strategies that could be more effective in reducing disparities. This thesis is mainly focused on building new statistical models for disparity studies in response to the complex data types which we encounter today. We propose tree-based models to unveil the distribution of disparities in a population through the hierarchical interaction between individual level variables (like clinical variables or genetic variables) and social determinants of health (like SES, education level etc). Precision medicine has the potential to revolutionize medicine because clinical decisions can in theory be made in a manner that is more customized to an individual patient. It’s not surprising then that there has been growing interest in trying to identify and reduce disparities using precision medicine constructs. Central to this paradigm is the search for what we term disparity subtypes. We will take the tree-based models we developed as well as another framework known as peeling and develop a new statistical framework for the identification of disparity subtypes. Even though much of disparity science has traditionally focused on social determinants of health, the move towards an integrative framework together with biological determinants requires that researchers must be able to find a common language and framework for connecting the two. Additionally, as biological data increases, so does the contextual information that researchers are collecting. Thus it’s imperative to be able to quantify
Inside-outside determination is a basic building block for higher-level geometry processing operations. Generalized winding numbers provide a robust answer for triangle meshes, regardless of defects such as self-inter...
详细信息
Inside-outside determination is a basic building block for higher-level geometry processing operations. Generalized winding numbers provide a robust answer for triangle meshes, regardless of defects such as self-intersections, holes or degeneracies. In this paper, we further generalize the winding number to point clouds. Previous methods for evaluating the winding number are slow for completely disconnected surfaces, such as triangle soups or-in the extreme case-point clouds. We propose a tree-based algorithm to reduce the asymptotic complexity of generalized winding number computation, while closely approximating the exact value. Armed with a fast evaluation, we demonstrate the winding number in a variety of new applications: voxelization, signing distances, generating 3D printer paths, defect-tolerant mesh booleans and point set surfaces.
tree-based ensemble algorithms (TEAs) have made significant advances in recent years due to their simple algorithmic design. However, when the proportion of the 'most informative' features is low, the performa...
详细信息
tree-based ensemble algorithms (TEAs) have made significant advances in recent years due to their simple algorithmic design. However, when the proportion of the 'most informative' features is low, the performance of conventional TEAs degrades significantly. The primary rationale for performance degradation is that traditional algorithmic design appears to be biased toward the least informative features, and the sub-space selection procedure contains uninformative features. This paper proposes a logically randomized forest (LRF) algorithm by incorporating two different enhancements into existing TEAs. The first enhancement is made to address the issue of biasness by performing feature-level engineering. The second enhancement is the approach by which individual feature sub-spaces are selected. To derive the first enhancement, we use the graph-theoretic principle of minimal vertex cover to construct a relevant assemblage of features. Following that, the permutation-based feature importance technique is employed to calculate the 'informativeness' of the relevant features in order to infuse logical randomness into the individual trees in the forest. For the second enhancement, the stratified sampling method is used to ensure that the most informative features are present in all newly created feature sub-spaces. Consequently, individual trees are generated using the Roulette wheel-based selection (RWS) algorithm. The proposed algorithm has been evaluated on two real -world genomic data sets, ten hybrid-synthetic classification data sets, and twenty multidisciplinary benchmark data sets with varying characteristics. The experimental findings demonstrate that the LRF outperforms the existing benchmark and cutting-edge TEAs.
The recent exponential growth in the data volume and number of identified pulsar stars is due to pulsar candidate search experiments and surveys. In this study, we investigated the existing methods and techniques used...
详细信息
The recent exponential growth in the data volume and number of identified pulsar stars is due to pulsar candidate search experiments and surveys. In this study, we investigated the existing methods and techniques used for pulsar prediction, such as applying filters based on pulsar observations, which can adversely affect the success of accurate pulsar prediction. Some of the existing methods are not capable of dealing with large volumes of data and others fail to accurately select the best candidates from pulsar observations. Thus, we developed a new approach based on the traditional supervised machine learning algorithm, which yields faster and more accurate results. In this study, we present our hybrid machine learning classifier called the random trees boosting voting classifier (RTB-VC) for predicting pulsar stars. RTB-VC combines tree-based classifiers and it employs the High Time Resolution Universe 2 (HTRU2) data set comprising a set of eight features related to pulsars and nonpulsars. The HTRU2 data set is imbalanced and we solve this problem by using the synthetic minority oversampling technique to generate artificial data and obtain a balanced data set. A feature set is used to separate pulsar and non-pulsar candidates because the different distributions of variables in the data set are helpful for training models. In the proposed approach, the prediction stage of RTB-VC is based on a combination of soft voting, hard voting, and weighted voting to obtain highly accurate and relevant criteria for finally predicting pulsars or non-pulsars. The ensemble-based structure of RTB-VC yields accurate results based on pulsar observations with a high F-1 score for pulsars (98.3%). We evaluated the learning algorithm in terms of its accuracy, precision, recall, and F-1 score. (C) 2020 Elsevier B.V. All rights reserved.
The Radio Frequency IDentification (RFID) is a wireless communication technology that enables automatic identification, tracking and data collection from any tagged object in a supply chain operating environment. A si...
详细信息
The Radio Frequency IDentification (RFID) is a wireless communication technology that enables automatic identification, tracking and data collection from any tagged object in a supply chain operating environment. A simple RFID system uses radio signals to transmit data via a tiny portable device, called a tag, which is read by an RFID reader and processed by the corporate information system (IS) to meet the needs of business management. One of the important performance issues in this system is to resolve RFID tag collision. Tag collision happens when two or more tags reflects-back their individual identification radio signals to the reader at the same time thus confusing the reader identification process. Different algorithmic solutions on tag collision are available. They are generally in two main categories of anti-collision problems: ALOHA-based solutions and tree-based solutions. However, ALOHA-basedalgorithms suffer from a time-related starvation problem. The well-known tree-based algorithms are Binary tree (BT) and Query tree (QT). In addition, QT algorithms are very efficient in memory utilization in comparison to other algorithmic solutions. This paper presents simulation-based experimental results on the performance of some well-known BT-basedalgorithms: simple Binary Search algorithm (BSA), Dynamic Binary Search algorithm (DBSA), and Backtrack Binary algorithm (BBA).
Along with the rapid development of Web of Things, the RFID technology is widely applied in every field, but today, the great challenge we face is how to avoid information conflict and collision in the process of acqu...
详细信息
Along with the rapid development of Web of Things, the RFID technology is widely applied in every field, but today, the great challenge we face is how to avoid information conflict and collision in the process of acquisition and treatment of massive information. It is the keystone of the study. This paper conducts a comparative analysis on different tree-based algorithms as improved, and integrated with multiple sub-cycle response mechanism, a Multi-Response Collision treealgorithm is proposed. And beyond that, this paper simulates and analyzed this algorithm and other improved ones. The results reveal that, compared to other algorithms, MRCT algorithm features a better performance, less recognition cycles, least query time slots on average, and ceiling throughput rate.
暂无评论