Digital Library[ Search Result ]
Performance Evaluation Technique of Learning Model Based on Feature Cluster in Sensing Data of Collaborative Robots
Jinse Kim, Subin Bea, Ye-Seul Park, Jung-Won Lee
http://doi.org/10.5626/JOK.2022.49.10.824
Recently, attempts have been made to apply an artificial intelligence model to PHM(Prognostics and Health Management) of collaborative robots, a representative equipment of smart factories. However, typical models are developed in a heuristic way without preprocessing or analysis of sensing data collected by operating test programs. Therefore, in this paper, we proposed a model performance evaluation method based on feature cluster concept which could analyze features of time series sensing data with cycles collected from cooperative robots. To demonstrate the effectiveness of the proposed method, we applied it to a program classification model, an internal component of the motion fault detection network, and identified characteristics of data that contributed to performance degradation, which has not been revealed by existing method. This results enabled a qualitative evaluation of the performance of the model and provided directions to improving model performance.
Analysis of Limits in Applying AP-QoS-based Wi-Fi Slicing for Real-Time Systems
Jin Hyun Kim, Hyonyoung Choi, Gangjin Kim, Yundo Choi, Tae-Won Ban, Se-Hoon Kim
http://doi.org/10.5626/JOK.2021.48.6.723
Network slicing is a new network technology that guarantees the quality of network services according to application services or user’s types. Wi-Fi, IEEE 802.11-based LAN, is the mostly popularly used short-range wireless network and has been continually attracting more and more from users. Recently, the use of Wi-Fi by safety critical IoT devices, such as medical devices, has been drastically increasing. Moreover, enterprises require network slicing of Wi-Fi to introduce the provision of prioritized QoS of Wi-Fi depending on the service type of customer. This paper presents the analysis of the limits and difficulties in applying AP-QoS-based network slicing for hard real-time systems that demand temporal deterministic streaming services. In this paper, we have defined a formal framework to analyze QoS-providing IEEE 802.11e Enhanced Distributed Coordination Access and provide the worst-case streaming scenarios and thereby demonstrated why the temporal determinism of network streaming is broken. In addition, simulation results of AP-QoS-based network slicing using NS-3 are presented to show the limits and difficulties of the network slicing. Moreover, we present Wi-Fi network slicing techniques based on EDCA of AP-QoS for real-time systems through our technical report referenced in this paper.
ILP-based Schedule Synthesis of Time-Sensitive Networking
Jin Hyun Kim, Hyonyoung Choi, Kyong Hoon Kim, Insup Lee, Se-Hoon Kim
http://doi.org/10.5626/JOK.2021.48.6.595
IEEE 802.1Qbv Time Sensitive Network (TSN), the latest real-time Ethernet standard, is a network designed to guarantee the temporal accuracy of streams. TSN is an Ethernet-based network system that is actively being developed for the factory automation and automobile network systems. TSN controls the flow of data streams based on schedules generated statically off-line to satisfy end-to-end delay or jitter requirements. However, the generation of TSN schedules is an NP-hard problem; because of this, constraint solving techniques, such as SMT (Satisfiability Modulo Theory) and ILP (Integer Linear Programming), have mainly been proposed as solutions to this problem. This paper presents a new approach using a heuristic greedy and incremental algorithm working with ILP to decrease the complexity of computing schedules and improve the schedule generation performance in computing TSN schedules. Finally, we compare our proposed method with the existing SMT solver approach to show the performance of our approach.
Effect of Denoising Autoencoder in the view of Item Popularity Bias
Jinhong Kim, Jae-woong Lee, Jongwuk Lee
http://doi.org/10.5626/JOK.2021.48.5.575
Denoising autoencoder (DAE) is commonly used in recent recommendation systems. It is a type of Autoencoder that trains by giving noise to the input and has shown improved performance compared to autoencoder. In this paper, we analyze the effect of noise in terms of item popularity to interpret the training of DAE. For analysis, we design the experiment in the following two ways. First, we observe the changes of the learned item vector’s L2-norm by giving noise to the autoencoder. Second, by giving noise only to presampled items by popularity, we anlayze whether the improved performance of the DAE is related to item popularity. Results of the experiment showed that the variance of the item vector norm caused by popularity was reduced by noise, and that the accuracy increased when noise was given to the popular items.
Facial Emotion Recognition Data Augmentation using Generative Adversarial Network
http://doi.org/10.5626/JOK.2021.48.4.398
The facial emotion recognition field of computer vision has recently been identified to demonstrate meaningful results through various neural networks. However, the major datasets of facial emotion recognition have the problem of “class imbalance,” which is a factor that degrades the accuracy of deep learning models. Therefore, numerous studies have been actively conducted to solve the problem of class imbalance. In this paper, we propose “RDGAN,” a facial emotion recognition data augmentation model that uses a GAN to solve the class imbalance of the FER2013 and RAF_single that are used as facial emotion recognition datasets. RDGAN is a network that generates images suitable for classes by adding expression discriminators based on the image-to-image translation model between the existing images as compared to the prevailing studies. The dataset augmented with RDGAN showed an average performance improvement of 4.805%p and 0.857%p in FER2013 and RAF_single, respectively, compared to the dataset without data augmentation.
Wave Celerity Estimation using Unsupervised Image Registration from Video Imagery
Jinah Kim, Jaeil Kim, Sungwon Shin
http://doi.org/10.5626/JOK.2019.46.12.1296
In this paper, we propose an image registration method based on unsupervised learning to estimate wave celerity by tracking wave movements using a large amount of video imagery. It is difficult to estimate the wave celerity accurately using physics-based modeling in the coastal region, owing to the limitations of in-situ measurement and the high nonlinearity of wave phenomena itself as well as high complexity from nonlinear interactions. In order to estimate wave celerity, the proposed method learns the nonlinear wave behavior from the video imagery. Autoencoder is applied to separate hydrodynamics scenes from environmental factors, such as daylights. The displacement vector of propagating waves is computed by non-linear spatio-temporal image registration. The wave celerity is estimated by accumulating the displacement vectors along time. In this paper, we compare the wave celerity measurement with conventional image processing methods and actual measurement using sensors for accuracy evaluation.
Deep Learning Model based on Autoencoder for Reducing Algorithmic Bias of Gender
http://doi.org/10.5626/JOK.2019.46.8.721
Algorithmic bias is a discrimination that is reflected in the model by a bias in data or combination of characteristics of model and data in the algorithm. In recent years, it has been identified that the bias is not only present but also amplified in the deep learning model; thus, there exists a problem related to bias elimination. In this paper, we analyze the bias of the algorithm by gender in terms of bias-variance dilemma and identify the cause of bias. To solve this problem, we propose a deep auto-encoder based latent space matching model. Based on the experimental results, it is apparent that the algorithm bias in deep learning is caused by difference of the latent space for each protected feature in the feature extraction part of the model. A model proposed in this paper achieves the low bias by reducing the differences in extracted features by transferring data with different gender characteristics to the same latent space. We employed Equality of Odds and Equality of Opportunity as a quantitative measure and proved that proposed model is less biased than the previous model. The ROC curve shows a decrease in the deviation of the predicted values between the genders.
Data-driven Path Selection for Improving Industrial-Strength Static Analyzers
http://doi.org/10.5626/JOK.2019.46.4.363
We propose a data-driven method to improve path-sensitive industrial-strength static analyzers. Most industrial static analyzers adopt path-sensitive techniques and path selection holds the key to their performance. We propose a method to automatically learn new cost-effective path-selection heuristics from an existing analyzer with a manually tuned path-selection heuristic. We evaluated our method on an industrial static C code bug-finder from Sparrow as a baseline analyzer with 17 C open-source benchmark programs. The experimental results showed that with the newly-learned path-selection heuristic, the analyzer reported 90.8% of the defects in only 38% of the analysis time, compared to the baseline analysis. This method reported more defects in less time than the baseline path-selection heuristic under similar path search space constraints.
Effective Generative Chatbot Model Trainable with a Small Dialogue Corpus
Jintae Kim, Hyeon-gu Lee, Harksoo Kim
http://doi.org/10.5626/JOK.2019.46.3.246
Contrary to popular retrieval-based chatbot models, generative chatbot models do not depend on predefined responses, but rather generate new responses based on well-trained neural networks. However, they require a large number of training corpus in the form of query-response pairs. If the training corpus are insufficient, they make grammatical errors emanating from out-of-vocabulary or sparse data problems, mostly in longer sentences. To overcome this challenge, we proposed a chatbot model based on sequence-to-sequence neural network using a mixture of words and syllables as encoding-decoding units. Moreover, we proposed a two-step training procedure involving pre-training using a large non-dialogue corpus and retraining using a smaller dialogue corpus. In the experiment involving small dialogue corpus (47,089 query-response pairs for training and 3,000 query-response pairs for evaluation), the proposed encoding-decoding units resulted to a reduction in out-of-vocabulary problem while the two-step training method led to improved performance measures like BLEU and ROUGE.
Document Summarization Using TextRank Based on Sentence Embedding
Seok-won Jeong, Jintae Kim, Harksoo Kim
http://doi.org/10.5626/JOK.2019.46.3.285
Document summarization is creating a short version document that maintains the main content of original document. An extractive summarization has been actively studied by the reason of it guarantees the basic level of grammar and high level of accuracy by copying a large amount of text from the original document. It is difficult to consider the meaning of sentences because the TextRank, which is a typical extractive summarization method, calculates an edge of graph through the frequency of words. In a bid to solve these drawbacks, we propose a new TextRank using sentence embedding. Through experiments, we confirmed that the proposed method can consider the meaning of the sentence better than the existing method.
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