Computing and Informatics (E-Journal - Institute of Informatics, SAS, Bratislava)
Not a member yet
1506 research outputs found
Sort by
Effective Utilization of Supervised Learning Techniques for Process Model Matching
Get PDFThe recent attempts to use supervised learning techniques for process model matching have yielded below par performance. To address this issue, we have transformed the well-known benchmark correspondences to a readily usable format for supervised learning. Furthermore, we have performed several experiments using eight supervised learning techniques to establish that imbalance in the datasets is the key reason for the abysmal performance. Finally, we have used four data balancing techniques to generate balanced training dataset and verify our solution by repeating the experiments for the four datasets, including the three benchmark datasets. The results show that the proposed approach increases the matching performance significantly
Travel Mode Recognition from GPS Data Based on LSTM
Get PDFA large amount of GPS data contains valuable hidden information. With GPS trajectory data, a Long Short-Term Memory model (LSTM) is used to identify passengers' travel modes, i.e., walking, riding buses, or driving cars. Moreover, the Quantum Genetic Algorithm (QGA) is used to optimize the LSTM model parameters, and the optimized model is used to identify the travel mode. Compared with the state-of-the-art studies, the contributions are: 1. We designed a method of data processing. We process the GPS data by pixelating, get grayscale images, and import them into the LSTM model. Finally, we use the QGA to optimize four parameters of the model, including the number of neurons and the number of hidden layers, the learning rate, and the number of iterations. LSTM is used as the classification method where QGA is adopted to optimize the parameters of the model. 2. Experimental results show that the proposed approach has higher accuracy than BP Neural Network, Random Forest and Convolutional Neural Networks (CNN), and the QGA parameter optimization method can further improve the recognition accuracy
Probabilistic Page Replacement Policy in Buffer Cache Management for Flash-Based Cloud Databases
Get PDFIn the fast evolution of storage systems, the newly emerged flash memory-based Solid State Drives (SSDs) are becoming an important part of the computer storage hierarchy. Amongst the several advantages of flash-based SSDs, high read performance, and low power consumption are of primary importance. Amongst its few disadvantages, its asymmetric I/O latencies for read, write and erase operations are the most crucial for overall performance. In this paper, we proposed two novel probabilistic adaptive algorithms that compute the future probability of reference based on recency, frequency, and periodicity of past page references. The page replacement is performed by considering the probability of reference of cached pages as well as asymmetric read-write-erase properties of flash devices. The experimental results show that our proposed method is successful in minimizing the performance overheads of flash-based systems as well as in maintaining the good hit ratio. The results also justify the utility of a genetic algorithm in maximizing the overall performance gains
Investigation of Parallel Data Processing Using Hybrid High Performance CPU + GPU Systems and CUDA Streams
Get PDFThe paper investigates parallel data processing in a hybrid CPU+GPU(s) system using multiple CUDA streams for overlapping communication and computations. This is crucial for efficient processing of data, in particular incoming data stream processing that would naturally be forwarded using multiple CUDA streams to GPUs. Performance is evaluated for various compute time to host-device communication time ratios, numbers of CUDA streams, for various numbers of threads managing computations on GPUs. Tests also reveal benefits of using CUDA MPS for overlapping communication and computations when using multiple processes. Furthermore, using standard memory allocation on a GPU and Unified Memory versions are compared, the latter including programmer added prefetching. Performance of a hybrid CPU+GPU version as well as scaling across multiple GPUs are demonstrated showing good speed-ups of the approach. Finally, the performance per power consumption of selected configurations are presented for various numbers of streams and various relative performances of GPUs and CPUs
Adaptive Fault Diagnosis of Motors Using Comprehensive Learning Particle Swarm Optimizer with Fuzzy Petri Net
Get PDFThis study proposes and applies a comprehensive learning particle swarm optimization (CLPSO) fuzzy Petri net (FPN) algorithm, which is based on the CLPSO algorithm and FPN, to the fault diagnosis of a complex motor. First, the transition confidence is replaced by a Gaussian function to deal with the uncertainty of fault propagation. Then, according to the Petri net principle, a competition operator is introduced to improve the matrix reasoning. Finally, a CLPSO-FPN model for motor fault diagnosis is established based on the motor failure mechanism and fault characteristics. The CLPSO algorithm is used to generate the system parameters for fault diagnosis and to improve the adaptability and accuracy of fault diagnosis. This study considers the example of a three-phase asynchronous motor. The results show that the proposed algorithm can diagnose faults in this motor with satisfactory adaptability and accuracy compared with the traditional FPN algorithm. By establishing the system model, the fault propagation process of motors can be accurately and intuitively expressed, thus improving the fault treatment and equipment maintenance of motors
Time-Sensitive Collaborative Filtering Algorithm with Feature Stability
Get PDFIn the recommendation system, the collaborative filtering algorithm is widely used. However, there are lots of problems which need to be solved in recommendation field, such as low precision, the long tail of items. In this paper, we design an algorithm called FSTS for solving the low precision and the long tail. We adopt stability variables and time-sensitive factors to solve the problem of user's interest drift, and improve the accuracy of prediction. Experiments show that, compared with Item-CF, the precision, the recall, the coverage and the popularity have been significantly improved by FSTS algorithm. At the same time, it can mine long tail items and alleviate the phenomenon of the long tail
PSVDAG: Compact Voxelized Representation of 3D Scenes Using Pointerless Sparse Voxel Directed Acyclic Graphs
Get PDFThis paper deals with the issue of geometry representation of voxelized three-dimensional scenes using hierarchical data structures. These include pointerless Sparse Voxel Octrees that have no pointers on child nodes and allow a compact binary representation. However, if necessary, there is a possibility to reconstruct these pointers for rapid traversing. Sparse Voxel Directed Acyclic Graphs added 32-bit pointers to child nodes and merging of common subtrees, which can be considered lossless compression. By merging common subtrees, no decompression overhead occurs at the time of traversing. The hierarchical data structure proposed herein - the Pointerless Sparse Voxel Directed Acyclic Graph - incorporates the benefits of both - pointerless Sparse Voxel Octrees (by avoiding storing pointers on child nodes) and Sparse Voxel Directed Acyclic Graphs (by allowing the merging of common subtrees due the introduction of labels and callers). The proposed data structure supports the quick and easy reconstruction of pointers by introducing the Active Child Node Count. It also potentially allows Child Node Mask compression of its nodes. This paper presents the proposed data structure and its binary-level encoding in detail. It compares the effectiveness of the representation of voxelized three-dimensional scenes (originally represented in OBJ format) in the proposed data structure with the data structures mentioned above. It also summarizes statistical data providing a more detailed description of the various parameters of the data structure for different scenes stored in multiple resolutions
A More Faithful Formal Definition of the Desired Property for Distributed Snapshot Algorithms to Model Check the Property
Get PDFThe first distributed snapshot algorithm was invented by Chandy and Lamport: Chandy-Lamport distributed snapshot algorithm (CLDSA). Distributed snapshot algorithms are crucial components to make distributed systems fault tolerant. Such algorithms are extremely important because many modern key software systems are in the form of distributed systems and should be fault tolerant. There are at least two desired properties such algorithms should satisfy: 1) the distributed snapshot reachability property (called the DSR property) and 2) the ability to run concurrently with, but not alter, an underlying distributed system (UDS). This paper identifies subtle errors in a paper on formalization of the DSR property and shows how to correct them. We give a more faithful formal definition of the DSR property; the definition involves two state machines - one state machine M_UDS that formalizes a UDS and the other M_CLDSA that formalizes the UDS on which CLDSA is superimposed (UDS-CLDSA) - and can be used to more precise model checking of the DSR property for CLDSA. We also prove a theorem on equivalence of our new definition and an existing one that only involves M_CLDSA to guarantee the validity of the existing model checking approach. Moreover, we prove the second property, namely that CLDSA does not alter the behaviors of UDS
Framework for Knowledge Discovery in Educational Video Repositories
Get PDFThe ease of creating digital content coupled with technological advancements allows institutions and organizations to further embrace distance learning. Teaching materials also receive attention, because it is difficult for the student to obtain adequate didactic material, being necessary a high effort and knowledge about the material and the repository. This work presents a framework that enables the automatic metadata generation for materials available in educational video repositories. Each module of the framework works autonomously and can be used in isolation, complemented by another technique or replaced by a more appropriate approach to the field of use, such as repositories with other types of media or other content
A Logic Petri Net-Based Repair Method of Process Models with Incomplete Choice and Concurrent Structures
Get PDFCurrent model repair methods cannot repair incomplete choice and concurrent structures precisely and simply. This paper presents a repair method of process models with incomplete choice and concurrent structures via logic Petri nets. The relation sets are constructed based on process trees, including branch sets, choice activity sets and concurrent activity sets. The deviations are determined by analyzing the relation between relation sets and activities in the optimal alignment. The model repair method is proposed for models with incomplete choice and concurrent structures via logic Petri nets according to different deviation positions. Finally, the correctness and effectiveness of the logic Petri net-based repair method are illustrated by simulation experiments