We review a new form of self-organizing map which is based on a nonlinear projection of latent points into data space, identical to that performed in the Generative Topographic Mapping (GTM) [1]. But whereas the GTM i...
详细信息
ISBN:
(纸本)9783540772255
We review a new form of self-organizing map which is based on a nonlinear projection of latent points into data space, identical to that performed in the Generative Topographic Mapping (GTM) [1]. But whereas the GTM is an extension of a mixture of experts, this model is an extension of a product of experts [6]. We show visualisation and clustering results on a data set composed of video data of lips uttering 5 Korean vowels and show that the new mapping achieves better results than the standard Self-Organizing Map.
this work investigates learning and generalisation capabilities of Radial Basis Function Networks used to solve function regression and classification tasks in the environmental context. In particular RBFN is applied ...
详细信息
ISBN:
(纸本)9783642153808
this work investigates learning and generalisation capabilities of Radial Basis Function Networks used to solve function regression and classification tasks in the environmental context. In particular RBFN is applied to solve the problem of snow cover thickness estimation in which critical aspects such as minimal training condition, weak pattern description and inconsistency among data arise. the RBFN shows good performances and high flexibility in coping with regression, hard and soft classifications which are complementary tasks in the analysis of complex environmental phenomena.
Developing a data warehouse for XML documents involves two major processes: one of creating it, by processing XML raw documents into a specified data warehouse repository;and the other of querying it, by applying tech...
详细信息
the threat environment is rapidly changing and the cyber security skill shortage is a widely acknowledged problem. However, teaching such skills and keeping professionals up-to-date is not trivial. New malware types a...
详细信息
ISBN:
(纸本)9783319911526;9783319911519
the threat environment is rapidly changing and the cyber security skill shortage is a widely acknowledged problem. However, teaching such skills and keeping professionals up-to-date is not trivial. New malware types appear daily, and it requires significant time and effort by a teacher to prepare a unique, current and challenging courses in the malware reverse engineering. Novel teaching methods and tools are required. this paper describes an experience with an automated hands-on learning environment in a malware reverse engineering class taught at Tallinn University of Technology in Estonia. Our hands-on practical lab is using a fully automated Cyber Defense Competition platform intelligent Training Exercise Environment (i-tee) [1] combined with typical Capture-the-Flag competition structure and open-source tools where possible. We describe the process of generating a unique and comparable reverse-engineering challenge and measuring the students' progress through the process of analysis, reporting flags and debugging data, recording and taking into account their unique approach to the task. We aim to measure the students' using the Bloom's taxonomy, i.e., mastering the art of malware reverse engineering at the higher cognitive levels. the presented teaching and assessment method builds foundation for enhancing the future malware reverse engineering training quality and impact.
Withthe continuous innovation of computer science as well as the big data acquisition technology, machine learning (ML), developed as a state-of-art framework, now has been comprehensively applied in speed prediction...
详细信息
ISBN:
(纸本)9781728194097
Withthe continuous innovation of computer science as well as the big data acquisition technology, machine learning (ML), developed as a state-of-art framework, now has been comprehensively applied in speed prediction tasks. However, ML methods usually require intensive hyper-parameter tuning, which hinders the practical deployment of ML models. In view of this, this paper proposes an automated machine learning (AutoML) framework for speed prediction, which enables the prediction work to be accomplished in a much more timesaving and convenient way as well as in high prediction accuracy. the proposed framework utilizes the Genetic Algorithm (GA) following its four major procedures: Genome coding, Crossover, Mutation and Selection to automatically search for the optimal neural network architectures and hyperparameters. the proposed framework is examined on a real-world large-scale dataset in the city of Berlin, Germany. the experimental results demonstrate that the proposed method outperforms other benchmarking methods by a significant margin. Sensitivity analysis is also conducted to show the robustness of the proposed method. this study demonstrates the great penitential of using AutoML in traffic speed prediction and other related transportation applications.
Because of the risky nature of stock market, most people do not feel a secure option to invest their money in financial trading. Focusing on this basic concern of investors much research efforts are devoted to develop...
详细信息
ISBN:
(纸本)9781728188676
Because of the risky nature of stock market, most people do not feel a secure option to invest their money in financial trading. Focusing on this basic concern of investors much research efforts are devoted to develop automated trading systems that make intelligent decisions according to the market situation and help investor to make profit beside risk. In contrast, in this paper we proposed multi-objective systems based on deep reinforcement learning for stock trading. Target of the multi-objective systems is to get maximum profit by adjusting risk. We design the whole structure of systems consisting two deep neural networks first is LSTM autoencoder for robust feature extraction and second deep reinforcement learning with LSTM recurrent neural network for decision making in order to achieve the investor's goal. We conduct an experiment on real historic data for verification of the systems and compare them with conventional trading systems.
this paper presents several improvements to the framework of information-preserving empirical mode decomposition (EMD). the basic framework was presented in our previous work [1]. the method decomposes a non-stationar...
详细信息
ISBN:
(纸本)9783642153808
this paper presents several improvements to the framework of information-preserving empirical mode decomposition (EMD). the basic framework was presented in our previous work [1]. the method decomposes a non-stationary neural response into a number of oscillatory modes varying in information content. After the spectral information analysis only few modes, taking part in stimulus coding, are retrieved for further analysis. the improvements and enhancement have been proposed for the steps involved in information quantification and modes extraction. An investigation has also been carried out for compression of retrieved informative modes of the neural signal in order to achieve a lower bit rate using the proposed framework. Experimental results are presented.
A Gaussian mixture model (GMM) estimates a probability density function using the expectation-maximization algorithm. However, it may lead to a poor performance or inconsistency. this paper analytically shows that per...
详细信息
In telecom industry high installation and marketing costs make it between six to ten times more expensive to acquire a new customer than it is to retain the existing one. Prediction and prevention of customer chum is ...
详细信息
ISBN:
(纸本)3540454853
In telecom industry high installation and marketing costs make it between six to ten times more expensive to acquire a new customer than it is to retain the existing one. Prediction and prevention of customer chum is therefore a key priority for industrial research. While all the motives of customer decision to churn are highly uncertain there is lots of related temporal data sequences generated as a result of customer interaction withthe service provider. Existing churn prediction methods like decision tree typically just classify customers into chumers or non-chumers while completely ignoring the timing of chum event. Given histories of other customers and the current customer's data, the presented model proposes a new k nearest sequence (kNS) algorithm along with temporal sequence fusion technique to predict the whole remaining customer data sequence path up to the chum event. It is experimentally demonstrated that the new model better exploits time-ordered customer data sequences and surpasses the existing churn prediction methods in terms of performance and offered capabilities.
the query optimization problem in data base and data warehouse management systems is quite similar. Changes to Joins sequences, projections and selections, usage of indexes, and aggregations are all decided during the...
详细信息
ISBN:
(纸本)9783642153808
the query optimization problem in data base and data warehouse management systems is quite similar. Changes to Joins sequences, projections and selections, usage of indexes, and aggregations are all decided during the analysis of an execution schedule. the main goal of these changes is to decrease the query response time. the optimization operation is often dedicated to a single node. this paper proposes optimization to grid or cluster data warehouses / databases. Tests were conducted in a multi-agent environment, and the optimization focused not only on a single node but on the whole system as well. A new idea is proposed here with multi-criteria optimization that is based on user-given parameters. Depending on query time, result admissible errors, and the level of system usage, task results were obtained along with grid optimization.
暂无评论