Analysis of experimental data has one of the most important roles in High Energy Physics. Commonly used multivariate techniques such as Boosted Decision Trees or Bayesian Neural Networks are based on learning algorith...
详细信息
Analysis of experimental data has one of the most important roles in High Energy Physics. Commonly used multivariate techniques such as Boosted Decision Trees or Bayesian Neural Networks are based on learning algorithms using Monte Carlo generated samples. We implemented a new Model Based Clustering method using Bayesian statistics and a modified iterative Expectation-Maximization algorithm for weighted data that have never been applied in this area. This greatly promising method was developed especially for the data collected from the Dempty set experiment, which was one of two large particle physics experiments at the Tevatron proton-antiproton collider at Fermilab. We optimized and tested the proposed method in the single top search using a data sample of 9.7 fb(-1) of integrated luminosity, which corresponds to the entire Run II Dempty set dataset.
The proceedings contain 295 papers. The topics discussed include: never-ending language learning;smart data - how you and i will exploit bigdata for personalized digital health and many other activities;addressing hu...
ISBN:
(纸本)9781479956654
The proceedings contain 295 papers. The topics discussed include: never-ending language learning;smart data - how you and i will exploit bigdata for personalized digital health and many other activities;addressing human bottlenecks in bigdata;BayesWipe: a multimodal system for data cleaning and consistent query answering on structured bigdata;metadata capital: simulating the predictive value of self-generated health information (SGHI);towards building and evaluating a personalized location-based recommender system;on the performance of MapReduce: a stochastic approach;PGMHD: a scalable probabilistic graphical model for massive hierarchical data problems;FusionFS: toward supporting data-intensive scientific applications on extreme-scale high-performance computing systems;and detecting and identifying system changes in the cloud via discovery by example.
Deep learning methods have recently made notable advances in the tasks of classification and representation learning. These tasks are important for brain imaging and neuroscience discovery, making the methods attracti...
详细信息
The development of IoT (Internet of Things) applications poses a number of scientific and technological challenges which stem from the characteristics of the IoT domain itself. These include the huge and increasing nu...
详细信息
The communicating material is a new paradigm of Internet of Things. It is designed to perform efficient product control and ensure an information continuum all along the product life cycle. Therefore, storage of life ...
详细信息
ISBN:
(纸本)9781479943579
The communicating material is a new paradigm of Internet of Things. It is designed to perform efficient product control and ensure an information continuum all along the product life cycle. Therefore, storage of life cycle information anddata dissemination in communicating materials are very important issues. This paper provides solutions for storing data on the material by systematic dissemination through integrated ultra-small wireless sensors nodes using counter-based broadcasting scheme, hop-counter and probabilistic storage mechanisms. Different algorithms are developed for non-localized and localized dissemination. The performances of our solutions are evaluated for non-segmented and then segmented data. Comparison results between storage mechanisms via simulation using Castalia/OMNeT++ show that probabilistic algorithm provides uniform and efficient data dissemination than hop-counter one for different density level.
The widespread adoption of cloud computing is having a big impact on the environment since the energy consumption of data centers and the resulting emissions are significantly increasing. Researchers and practitioners...
详细信息
ISBN:
(纸本)9789462520226
The widespread adoption of cloud computing is having a big impact on the environment since the energy consumption of data centers and the resulting emissions are significantly increasing. Researchers and practitioners in this field are looking for methods to improve the energy efficiency of data centers and increase the use of green energy sources. In fact, besides the energy consumption, the greenness of a data center can be characterized by the quantity of CO2 emissions associated with the use of electricity (from a specific energy mix) and/or fuels (e.g., for heating or cooling). In this paper, we propose an approach in which environmental impacts are considered as an important factor for the selection of the cloud site for the deployment of applications. In detail, considering a user perspective and focusing on the assessment of energy consumption and CO2 emissions, this paper proposes a method to support the users towards greener choices in the deployment of cloud applications.
The network economy era brings new challenges to all aspects of society, and the construction of network culture industry also brings an urgent request. Due to the complexity of network culture industry, the construct...
详细信息
Some aspects of the theoretical and methodological basis for the use of remote sensing data of snow cover for hazards assessment related to meteorological, climatic, hydrological and hydrogeological risks over urban a...
详细信息
The routing algorithm of the DTN, which experience frequent long-duration partitions, is quite different from the normal networks. Active routing algorithms usually adopt an active host (data mule/message ferry) to ex...
详细信息
ISBN:
(纸本)9781849199285
The routing algorithm of the DTN, which experience frequent long-duration partitions, is quite different from the normal networks. Active routing algorithms usually adopt an active host (data mule/message ferry) to exchange and forward the messages. In the existing studies in DTN, the active routing problem is generally solved by adopting the solutions of the Travelling Salesman Problem (TSP). In such solutions, the active routing problem is often solved by finding a simple cycle with no repeated nodes nor edges. By considering a more general scenario, we propose an active routing algorithm, which is based on the historical information and node states, where the route is a closed walk. The simulations verify the accuracy and efficiency of the active routing algorithm we proposed in this paper.
The proceedings contain 10 papers. The topics discussed include: real time unobtrusive human behaviors recognition;long-term tracking in batmon: lessons and open challenges;anomaly detection using autoencoders with no...
ISBN:
(纸本)9781450331593
The proceedings contain 10 papers. The topics discussed include: real time unobtrusive human behaviors recognition;long-term tracking in batmon: lessons and open challenges;anomaly detection using autoencoders with nonlinear dimensionality reduction;spatial-temporal pyramid matching for crowd scene analysis;multi-action recognition via stochastic modeling of optical flow and gradients;multi-feature fusion based non negative matrix factorization: facial expression recognition from imaging sensors;multi-level analysis of peace and conflict data in GDELT;extracting user interests from graph connections for machine learning in location-based social networks;air pollution exposure estimation and finding association with human activity using wearable sensor network;and distributed feature selection with big sensor data.
暂无评论