1 |
Exploring mobile device usage patterns by using the FANN neural network libraryNémeth Norrby, Otto January 2016 (has links)
Security awareness is becoming an increasingly valuable characteristic due to the increased digitization of society. The commonality of constantly connected devices, such as smartphones and tablets, along with the threat of malware and cyber-attacks has sparked an interest in creating a system with the purpose of training people in security awareness. This thesis aims to show the presence of patterns in mobile device usage, and explore the possibility of using pattern detection as a means to predict riskful actions on mobile devices as a step to evaluate the prediction approach for use in the training system.A survey has been conducted by gathering usage data from a number of participants through the use of a logging application. This data was then analyzed using artificial neural networks provided by the open source FANN library in search for patterns preluding certain events. The results lend support to the claim that patterns exist in the way mobile devices are used, but the usefulness of FANN as a tool for finding these patterns was shown to be questionable.
|
2 |
Spatio-temporal Event Prediction via Deep Point Processes / 深層点過程を用いた時空間イベント予測Okawa, Maya 23 March 2022 (has links)
京都大学 / 新制・課程博士 / 博士(情報学) / 甲第24028号 / 情博第784号 / 新制||情||133(附属図書館) / 京都大学大学院情報学研究科知能情報学専攻 / (主査)教授 鹿島 久嗣, 教授 山本 章博, 教授 吉川 正俊 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DFAM
|
3 |
Studies on Fundamental Problems in Event-Level Language Analysis / イベントレベルの言語解析における基礎的課題に関する研究Kiyomaru, Hirokazu 23 March 2022 (has links)
京都大学 / 新制・課程博士 / 博士(情報学) / 甲第24029号 / 情博第785号 / 新制||情||133(附属図書館) / 京都大学大学院情報学研究科知能情報学専攻 / (主査)教授 黒橋 禎夫, 教授 河原 達也, 教授 鹿島 久嗣 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DFAM
|
4 |
Identification of topological and dynamic properties of biological networks through diverse types of dataGuner, Ugur 23 May 2011 (has links)
It is becoming increasingly important to understand biological networks in order to understand complex diseases, identify novel, safer protein targets for therapies and design efficient drugs. 'Systems biology' has emerged as a discipline to uncover biological networks through genomic data. Computational methods for identifying these networks become immensely important and have been growing in number in parallel to increasing amount of genomic data under the discipline of 'Systems Biology'.
In this thesis we introduced novel computational methods for identifying topological and dynamic properties of biological networks. Biological data is available in various forms. Experimental data on the interactions between biological components provides a connectivity map of the system as a network of interactions and time series or steady state experiments on concentrations or activity levels of biological constituents will give a dynamic picture of the web of these interactions. Biological data is scarce usually relative to the number of components in the networks and subject to high levels of noise. The data is available from various resources however it can have missing information and inconsistencies. Hence it is critical to design intelligent computational methods that can incorporate data from different resources while considering noise component.
This thesis is organized as follows; Chapter 1 and 2 will introduce the basic concepts for biological network types. Chapter 2 will give a background on biochemical network identification data types and computational approaches for reverse engineering of these networks. Chapter 3 will introduce our novel constrained total least squares approach for recovering network topology and dynamics through noisy measurements. We proved our method to be superior over existing reverse engineering methods. Chapter 4 is an extension of chapter 3 where a Bayesian parameter estimation algorithm is presented that is capable of incorporating noisy time series and prior information for the connectivity of network. The quality of prior information is critical to be able to infer dynamics of the networks. The major drawback of prior connectivity data is the presence of false negatives, missing links. Hence, powerful link prediction methods are necessary to be able to identify missing links. At this junction a novel link prediction method is introduced in Chapter 5. This method is capable of predicting missing links in a connectivity data. An application of this method on protein-protein association data from a literature mining database will be demonstrated. In chapter 6 a further extension into link prediction applications will be given. An interesting application of these methods is the drug adverse effect prediction. Adverse effects are the major reason for the failure of drugs in pharmaceutical industry, therefore it is very important to identify potential toxicity risks in the early drug development process. Motivated by this chapter 6 introduces our computational framework that integrates drug-target, drug-side effect, pathway-target and mouse phenotype-mouse genes data to predict side effects. Chapter 7 will give the significant findings and overall achievements of the thesis. Subsequent steps will be suggested that can follow the work presented here to improve network prediction methods.
|
5 |
Flood forecasting using time series data miningDamle, Chaitanya 01 June 2005 (has links)
Earthquakes, floods, rainfall represent a class of nonlinear systems termed chaotic, in which the relationships between variables in a system are dynamic and disproportionate, however completely deterministic. Classical linear time series models have proved inadequate in analysis and prediction of complex geophysical phenomena. Nonlinear approaches such as Artificial Neural Networks, Hidden Markov Models and Nonlinear Prediction are useful in forecasting of daily discharge values in a river. The focus of these methods is on forecasting magnitudes of future discharge values and not the prediction of floods. Chaos theory provides a structured explanation for irregular behavior and anomalies in systems that are not inherently stochastic. Time Series Data Mining methodology combines chaos theory and data mining to characterize and predict complex, nonperiodic and chaotic time series. Time Series Data Mining focuses on the prediction of events.
|
6 |
Análise, modelagem e predição de eventos em praças digitaisTrindade, Andrea Garcia January 2018 (has links)
Orientador: Prof. Dr. Carlos Alberto Kamienski / Dissertação (mestrado) - Universidade Federal do ABC, Programa de Pós-Graduação em Engenharia da Informação, Santo André, 2018. / A utilização das tecnologias de informação e comunicação (TICs) está cada dia mais presente na vida das pessoas, gerando mudanças em nosso comportamento, no modo como nos relacionamos, como nos comunicamos, ou ainda na forma como passamos a realizar certas ações. A cidade de São Paulo oferece desde 2014 acesso à Internet gratuitamente em 120 locais públicos através do Programa WiFiLivreSP. Embora estejam disponíveis dados de desempenho dessas praças, existe uma carência de informações sobre o perfil de utilização cotidiano destes locais e o impacto causado por ocorrência de eventos, elevando o número de usuários conectados. Este estudo modela estes perfis, analisando seu uso e procura realizar através da quantidade de usuários conectados a predição de situações atípicas (anomalias), caracterizando possíveis eventos e suas consequências. Para a modelagem e predição foram utilizadas as técnicas SARIMA e Suavização Exponencial Holt-Winters, onde a modelagem SARIMA demonstrou melhor desempenho e a detecção de possíveis anormalidades desenvolvida com a técnica de janelas deslizantes. Este estudo tem por finalidade desenvolver um conjunto de informações à tomada de decisões sobre as praças digitais já implantadas e parâmetros para análises em futuras implantações. / The use of information and communication technologies (ICTs) are increasingly present in people's lives, generating changes in our behavior, without terms how we relate, how we do not communicate, and how we do certain actions. Since 2014, the city of São Paulo has been offering free internet access on 120 public sites through WiFiLivreSP Project. Available performance data is available there is information about the daily usage profile of your sites and the impact caused by the occurrence of events, increasing the number of connected users. This study models these profiles, analyzing their use and seeks to perform through the number of users connected to a prediction of atypical situations (anomalies), characterizing events and their consequences. For a modeling and prediction with the SARIMA and Holt-Winters Exponential Smoothing techniques, where a SARIMA modeling demonstrated better performance and a detection of possible abnormalities developed with the sliding window technique. The purpose of this study is to develop a set of information to make decisions about how digital places already deployed and for analysis in future deployments.
|
7 |
Learning from multi-modal spatiotemporal data: machine learning approaches to advance resilience in smart gridsAlqudah, Mohammad, 0000-0001-7011-3762 12 1900 (has links)
The electric grid has been expanding both in size and the technologies used. As of the 2020s, the United States power grid consists of more than 9,200 electric generating units with more than 1 million megawatts of generating capacity connected to more than 300,000 miles of transmission lines. The United States electricity grid has rapidly expanded in recent decades, and the majority (over 70\%) of its infrastructure has exceeded 25 years of age. Due to its size and age, several challenges have emerged. Widespread power outages have been increasing across the United States. Between 2018 and 2020, more than 231,000 power outages occurred in the United States that lasted more than one hour, out of which 17,484 lasted at least eight hours. In the same period, the power outages resulted in an annual loss of 520 million customer hours across 2,447 U.S. counties. Moreover, and with the rapidly changing climate, between 2000 and 2021, approximately 83\% of significant power outages impacting a minimum of 50,000 customers in the United States were attributed to severe weather conditions. Lastly, the increasing use of renewables and other non-traditional generation methods forces the power system towards a more decentralized model, with many integrated systems constantly added to the grid. This decentralization adds additional burdens on controlling systems and grid operators. The rapid growth of technology and data storage allowed the deployment of sensing devices across the electric grid. Such technologies present a golden opportunity to tackle many of the electric grid's challenges. Despite that, such technologies presented many challenges simultaneously. With the large amounts of data, it became humanly impossible to comprehend, analyze, and use all collected data manually. While machine learning can be used to analyze smart grid data, this can be challenged by the nature of its data. Smart grid produces high-dimensional spatiotemporal data, and many applications require multi-modal data. Moreover, power systems' data quality challenges add complexities to model development. The data is noisy, contains missing segments, and usually has incomplete and inaccurate labels. In addition, interpreting machine learning models in the context of smart grids poses unique challenges. To address these challenges, different models for multiple smart-grid applications were introduced in this research, where each model focused on producing practical solutions for the challenges facing current-day smart grids. Using spatiotemporal data, a solar generation prediction model was proposed. The solution combined spatial and temporal data, then utilized machine learning embeddings to build datasets to train downstream models. This resulted in accurate prediction of solar generation across several settings. In addition to solar generation prediction, several models were introduced to detect, predict and explain power grid faults. A neural model is introduced to detect power faults from Phasor Measurement Unit (PMU) data. A novel method is introduced to preprocesses, de-noise, and combine high dimensional data, then this data is used to train novel neural methods that detect faults in multiple settings. This model addressed issues of high dimensionality and data quality. After that, several models studying power fault prediction and precursor discovery were introduced. A model that jointly predicts outages 6 hours ahead and produces explainable event precursors from multi-modal data is introduced. Where such precursors can assist power grid operators to take action to mitigate widespread power outages. Finally, a novel methodology is introduced that expands to previous work by predicting and extracting event precursors spatiotemporally 12 hours in advance. Where event precursors can be predicted on multiple spatial locations simultaneously, extracted spatiotemporal event precursors can help grid operators narrow down mitigation plans and help reduce the risk of widespread power outages. / Computer and Information Science
|
8 |
Investigation of Event-Prediction in Time-Series Data : How to organize and process time-series data for event prediction?Pradhan, Shameer Kumar January 2019 (has links)
The thesis determines the type of deep learning algorithms to compare for a particular dataset that contains time-series data. The research method includes study of multiple literatures and conduction of 12 tests. It deals with the organization and processing of the data so as to prepare the data for prediction of an event in the time-series. It also includes the explanation of the algorithms selected. Similarly, it provides a detailed description of the steps taken for classification and prediction of the event. It includes the conduction of multiple tests for varied timeframe in order to compare which algorithm provides better results in different timeframes. The comparison between the selected two deep learning algorithms identified that for shorter timeframes Convolutional Neural Networks performs better and for longer timeframes Recurrent Neural Networks has higher accuracy in the provided dataset. Furthermore, it discusses possible improvements that can be made to the experiments and the research as a whole.
|
9 |
The information content of options data applied to the prediction of clinical trial resultsYarger, Stephen A., 1974- 01 August 2011 (has links)
FDA decisions and late-stage clinical trial results regarding new pharmaceutical approvals can cause extreme moves in the share price of small biopharmaceutical companies. Throughout the clinical trial process, many potential investors are exposed to market-moving information before such information is made available to the investing public. An investor who wished to profit from advance knowledge about clinical trial results may use the publicly traded options markets in order to increase leverage and maximize profits.
This research examined options data surrounding the public release of information pertaining to the efficacy of clinical trials and approval decisions made by the FDA. Events were identified for small pharmaceutical companies with fewer than three currently approved drugs in an attempt to isolate the effect of individual clinical trial and FDA-related events on the share price of the underlying company. Option data were analyzed using logistic regression models in an attempt to predict phase II and III clinical trial outcome results and FDA new drug approval decisions. Implied volatility, open interest, and option contract delta values were the primary independent variables used to predict positive or negative event outcomes.
The dichotomized version of a predictor variable designed to estimate total investment exposure incorporating open interest, option contract delta values, and the underlying stock price was a significant predictor of negative pharmaceutical related events. However, none of
ii
the variables examined in this research were significant predictors of positive drug research related events.
The estimated total investment exposure variable used in this research can be applied to the prediction of future clinical trial and FDA decision related events when this predictor variable shows a negative signal. Additional research would help confirm this finding by increasing the sample size of events that potentially follow the same pattern as those examined in this research. / text
|
10 |
Fusion of Soft and Hard Data for Event Prediction and State EstimationThirumalaisamy, Abirami 11 1900 (has links)
Social networking sites such as Twitter, Facebook and Flickr play an important role
in disseminating breaking news about natural disasters, terrorist attacks and other
events. They serve as sources of first-hand information to deliver instantaneous news
to the masses, since millions of users visit these sites to post and read news items regularly.
Hence, by exploring e fficient mathematical techniques like Dempster-Shafer
theory and Modi ed Dempster's rule of combination, we can process large amounts of
data from these sites to extract useful information in a timely manner. In surveillance
related applications, the objective of processing voluminous social network data is to
predict events like revolutions and terrorist attacks before they unfold. By fusing the
soft and often unreliable data from these sites with hard and more reliable data from
sensors like radar and the Automatic Identi cation System (AIS), we can improve
our event prediction capability. In this paper, we present a class of algorithms to
fuse hard sensor data with soft social network data (tweets) in an e ffective manner.
Preliminary results using are also presented. / Thesis / Master of Applied Science (MASc)
|
Page generated in 0.172 seconds