Vol. 45, No. 5,
May 2018
Digital Library
Device Displacement Detection Technique Using Sound Properties and TDoA
http://doi.org/10.5626/JOK.2018.45.5.409
Devices without IMU (Inertial Measurement Unit) sensors such as notebooks cannot implement anti-theft applications. This is because there is no way to know the physical state of the devices at the S/W level. In this paper, we focus on the fact that speakers and a microphones are installed in every notebook. We then, use a sound characteristic and an analysis technique to suggest a method of detecting the displacement of a device. For example, if you leave your laptop in the library for a while, the technique suggested in this paper will detect if someone raises the laptop placed on your desk and use it to prevent theft. This technique can be applied to any device that is equipped with speakers and microphones. Therefore, it can be implemented in another object that needs to detect displacement, such as electronic devices as well as museum exhibits and frames.
Software Black Box
http://doi.org/10.5626/JOK.2018.45.5.416
Deterministic replay mechanisms have proved to be useful in many areas including debugging, fault tolerance, security, and postmortem analysis because they can deterministically reproduce a computer system’s execution. However, proposed full-system replay mechanisms have limited applicability because of their reliance on special hardware instrumentation or virtual machine (VM) technology. In this paper, we present a purely software-based approach to full-system replay, a software black box (SBB) that does not require either special hardware instrumentation or virtual machine technology. Our proposed SBB can deterministically replay a full software system, including both applications and the OS itself. We have implemented a prototype of SBB in an embedded RTOS on top of ARMv7 Cortex-A9 and have carried out experiments to evaluate our approach. Our experiments demonstrate that SBB can successfully reproduce subtle concurrency bugs, such as races and deadlocks that may occur both in applications and in the OS kernel. We also show that the event and data logging of SBB incurs such small performance overhead that it can be enabled permanently in the OS kernel.
VNSIM: Virtual Machine based Multi-core SSD Simulator for supporting NVM Express
http://doi.org/10.5626/JOK.2018.45.5.427
Solid State Drives (SSD) continue to improve its performance and capacity through the adoption of new host interfaces and the use of multi-channel/multi-way I/O parallelism with multiple core controllers. In order to design and evaluate the structure of the SSDs, a new SSD simulator needs to be developed that supports the latest storage techniques. In this study, we develop a SSD simulator, the Virtual-machine based NVMe SSD SIMulator (VNSIM), which supports the latest host controller interface, NVM Express. The VNSIM simulates the entire I/O stack, from applications to Flash memories. Unlike the existing SSD simulators, the VNSIM provides an environment for simulating and evaluating SSD structures with two or more Flash Translation Layer (FTL) cores running in the SSD. We developed the Flash I/O emulator which simulates the I/O performance of the Flash memory including page cache registers. The VNSIM was validated using the Samsung 950 Pro NVMe SSD, showing that the VNSIM models the 950 Pro SSD with a 6.2%～8.9% offset.
Morpheme-based Efficient Korean Word Embedding
Dongjun Lee, Yubin Lim, Ted “Taekyoung” Kwon
http://doi.org/10.5626/JOK.2018.45.5.444
Previous word embedding models such as word2vec and Glove are not able to learn the internal structure of words. This is a serious limitation for agglutinative languages with morphology such as Korean. In this paper, we propose a new model which is an expansion of the previous skip-gram model. This defines each word vector as a sum of its morpheme vectors and hence, learns the vectors of morphemes. To test the efficiency of our embedding, we conducted a word similarity test and a word analogy test. Furthermore, using our trained vectors on other NLP tasks, we tested how much performance actually had been enhanced.
Compression of Korean Phrase Structure Parsing Model using Knowledge Distillation
http://doi.org/10.5626/JOK.2018.45.5.451
A sequence-to-sequence model is an end-to-end model that transforms an input sequence into an output sequence of different lengths. However, it is difficult to apply to an actual service by using techniques such as attention mechanism and input-feeding to achieve high performance. In this paper, we apply the sequence-level knowledge distillation for natural language processing to the Korean phrase structure parsing, which is an effective technique for compressing the model. Experimental results show that when the size of the hidden layer is decreased from 500 to 50, the performance of F1 0.56% is improved and the speed is 60.71 times faster than that of the baseline model.
Effective Korean Token Units for Sequence Encoding in Deep Learning
http://doi.org/10.5626/JOK.2018.45.5.457
Deep learning has emerged as a new area of machine-learning research and has been successfully applied to natural language processing, such as machine translation and sentence classification. In this work, we use effective Korean input token units to encode Korean sentences for classification problems, such as topic detection. Recurrent and convolutional neural networks for Korean sentence encoding are briefly reviewed, and various Korean input tokens units, including character, morpheme-tag, morpheme, word, subword, syllable window, and hybrids of morpheme and character methods are explored. Extensive experiments on sentimental analysis, topic detection, and intention understanding tasks are conducted to find effective input token units.
An Approach to a Learning Prediction Model for Recognition of Daily Life Pattern based on Event Calculus
Seok-Hyun Bae, Sung-hyuk Bang, Hyun-Kyu Park, Myung-Joong Jeon, Je-Min Kim, Young-Tack Park
http://doi.org/10.5626/JOK.2018.45.5.466
Several studies have been conducted on data analysis and predicting results with the advance of machine learning algorithms. Still, there are many problems of cleaning the noise of the real-life dataset, which is disturbing a clear recognition on complex patterns of human intention. To overcome this limitation, this paper proposes an event calculus methodology with 3 additional steps for the recognition of human intention: intention reasoning, conflict resolution, and noise reduction. Intention reasoning identifies the intention of the living pattern time-series data. In conflict resolution, existing ongoing intentions and inferred intention are checked by a conflict graph, so that the intentions that can occur in parallel are inferred. Finally, for noise reduction, the inferred intention from the noise of living pattern data is filtered by the history of fluent. For the evaluation of the event calculus module, this paper also proposes data generation methodology based on a gaussian mixture model and heuristic rules. The performance estimation was conducted with 300 sequential instances with 5 intentions that were observed for 13 hours. An accuracy of 89.3% was achieved between the probabilistic model and event calculus module.
A MapReduce-based Prior Probability Optimization Algorithm for Topic Extraction
http://doi.org/10.5626/JOK.2018.45.5.478
Various topic extraction algorithms have been used to obtain meaningful information from a large number of text documents. Since the topic extraction algorithms work based on the Bayesian probability model, the prior probabilities, α and β, should be given as inputs. Until now, in order to run the topic extraction models, users have to either take advantage of default prior probability values or determine them subjectively. In this study, we propose a MapReduce-based prior probability optimization algorithm that systematically determines the prior probability values in addition to the improvement of performance and accuracy against a large-scale input data. Unlike the previous single thread algorithm, the proposed MapReduce-based algorithm quickly determines the prior probability values that are suitable for the input data. It then extracts topics with high accuracy after the topic extraction algorithm is executed with the chosen prior probability values. Our experimental results showed that the proposed method outperforms the previous method in the aspect of topic coherence and performance.
Efficient Ways of Attack for Network Isolation
Kyu Seok Han, Jiwon Yoon, Taekyu Kim, Young Woo Park, Jungkyu Han
http://doi.org/10.5626/JOK.2018.45.5.489
Many devices and objects have recently been connected to the network using the Internet of Thing technology. In a local area network (LAN) network for small scale, many devices are connected and the complexity of the network topology is greatly increased. Large-scale networks of such small-scale networks are also expanding nationwide. he flow of gathering and spreading data in a concentrated or distributed manner within a large network is being made. This is useful for various industries, financial, telecommunications, military, and power generation facilities in statebased industries use the nationwide Internet network to control and maintain a stream of data that can cope with emergency situations. In a network environment that has such a circumstance, if a critical device (node) or a small range of network (LAN) that is involved in the control, data collection, storage, or data processing is isolated and isolated from the entire network. This paper discusses techniques for isolating critical LANs or Nodes in large networks.
Global Discovery Service for Enhancing Performance of Intra- and Inter-Discovery Services in the Internet of Things
Kiwoong Kwon, Dongsoo Kim, Wondeuk Yoon, Daeyoung Kim
http://doi.org/10.5626/JOK.2018.45.5.502
The smart things on the Internet of Things (IoT) are connected to each other to generate a huge amount of thing data. This data is being stored in globally distributed repositories and used in a variety of IoT applications. Therefore, it is essential to access and retrieve a source of data repositories where the thing data is stored. GS1, an international nonprofit organization, was the first to propose the Discovery Service (DS), which searches for thing data repositories associated with the given thing. However, the existing DS studies focus on improving the performance of Inter-DS, which finds a specific one among distributed Intra-DS, so that the performance degradation of the Intra-DS may cause the degradation of the overall DS performance. To alleviate this problem, we propose the Global-DS (GDS) that considers both Inter- and Intra-DS performance. For performance evaluation, we constructed an experimental testbed, and tested the throughput and delay of GDS. The results showed that data partitioning, load balancing and caching improve the performance of GDS.
Search
Journal of KIISE
- ISSN : 2383-630X(Print)
- ISSN : 2383-6296(Electronic)
- KCI Accredited Journal
Editorial Office
- Tel. +82-2-588-9240
- Fax. +82-2-521-1352
- E-mail. chwoo@kiise.or.kr