• Journal of Information Processing Systems
• /
• v.16 no.4
• /
• pp.809-819
• /
• 2020

Gender prediction accuracy increases as convolutional neural network (CNN) architecture evolves. This paper compares voting and ensemble schemes to utilize the already trained five CNN models to further improve gender prediction accuracy. The majority voting usually requires odd-numbered models while the proposed softmax-based voting can utilize any number of models to improve accuracy. The ensemble of CNN models combined with one more fully-connected layer requires further tuning or training of the models combined. With experiments, it is observed that the voting or ensemble of CNN models leads to further improvement of gender prediction accuracy and that especially softmax-based voters always show better gender prediction accuracy than majority voters. Also, compared with softmax-based voters, ensemble models show a slightly better or similar accuracy with added training of the combined CNN models. Softmax-based voting can be a fast and efficient way to get better accuracy without further training since the selection of the top accuracy models among available CNN pre-trained models usually leads to similar accuracy to that of the corresponding ensemble models.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2019.01a
• /
• pp.37-38
• /
• 2019

본 논문에서는 자원제한적인 IoT 환경에서 스케줄링 정확도 향상을 위해 Softmax를 이용한 Q-learning 기반의 패킷 스케줄링 기법을 제안한다. 기존 Q-learning의 Exploitation과 Exploration의 균형을 유지하기 위해 e-greedy 기법이 자주 사용되지만, e-greedy는 Exploration 과정에서 최악의 행동이 선택될 수도 있는 문제가 발생한다. 이러한 문제점을 해결하기 위해 본 연구에서는 Softmax를 기반으로 다중 센서 노드 환경에서 데이터 패킷에 대한 Quality of Service (QoS) requirement 정확도를 높이기 위한 연구를 진행한다. 이 때 Temperature 매개변수를 사용하는데, 이는 새로운 정책을 Explore 하기 위한 매개변수이다. 본 논문에서는 시뮬레이션을 통하여 제안된 Softmax를 이용한 Q-learning 기반의 패킷 스케줄링 기법이 기존의 e-greedy를 이용한 Q-learning 기법에 비해 스케줄링 정확도 측면에서 우수함을 보인다.

• Convergence Security Journal
• /
• v.20 no.4
• /
• pp.81-90
• /
• 2020

In the real world, new types of attacks or variants are constantly emerging, but attack traffic classification models developed through artificial neural networks and supervised learning do not properly detect new types of attacks that have not been trained. Most of the previous studies overlooked this problem and focused only on improving the structure of their artificial neural networks. As a result, a number of new attacks were frequently classified as normal traffic, and attack traffic classification performance was severly degraded. On the other hand, the softmax function, which outputs the probability that each class is correctly classified in the multi-class classification as a result, also has a significant impact on the classification performance because it fails to calculate the softmax score properly for a new type of attack traffic that has not been trained. In this paper, based on this characteristic of softmax function, we propose an efficient method to improve the classification performance against new types of attacks by classifying traffic with a probability below a certain level as attacks, and demonstrate the efficiency of our approach through experiments.

• Annual Conference of KIPS
• /
• 2021.11a
• /
• pp.584-587
• /
• 2021

일반적으로 얼굴인식 시스템은 영상에서 추출한 Feature와 DB 상의 Feature를 비교하는 구조를 가지고 있다. 하지만 원하는 Class의 Feature만 보고 DB 상에서 일치하는 Class의 위치를 특정하는 것은 불가능하기에 DB 상의 모든 Feature와 비교하는 절차가 필요하다. DB 크기가 커짐에 따라 처리시간과 메모리상의 문제가 발생하는데, 이 논문에서는 이를 해결하기 위한 Deep Hashing 모델을 제안한다. Softmax 기반의 Loss를 이용하여 학습하였고, 8-bits의 해시를 추출하였을 때 53%의 Feature 일치율을 보였으며, 이를 사용할 경우 DB 평균 대조군을 23% 이하로 줄이는 효과를 볼 수 있을 것으로 추정한다.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.23 no.1
• /
• pp.27-32
• /
• 2019

Research that implements the classification process through Deep Learning algorithm, one of the outstanding human abilities, includes a unimodal model, a multi-modal model, and a multi-modal method using music videos. In this study, the results were better by suggesting a system to analyze each song's spectrum into short samples and vote for the results. Among Deep Learning algorithms, CNN showed superior performance in the category of music genre compared to RNN, and improved performance when CNN and RNN were applied together. The system of voting for each CNN result by Deep Learning a short sample of music showed better results than the previous model and the model with Softmax layer added to the model performed best. The need for the explosive growth of digital media and the automatic classification of music genres in numerous streaming services is increasing. Future research will need to reduce the proportion of undifferentiated songs and develop algorithms for the last category classification of undivided songs.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.67 no.1
• /
• pp.114-123
• /
• 2018

This paper concerns a design and learning method of softmax function neural networks based on K-means clustering. The partial discharge data Information is preliminarily processed through simulation using an Epoxy Mica Coupling sensor and an internal Phase Resolved Partial Discharge Analysis algorithm. The obtained information is processed according to the characteristics of the pattern using a Motor Insulation Monitoring System program. At this time, the processed data are total 4 types that void discharge, corona discharge, surface discharge and slot discharge. The partial discharge data with high dimensional input variables are secondarily processed by principal component analysis method and reduced with keeping the characteristics of pattern as low dimensional input variables. And therefore, the pattern classifier processing speed exhibits improved effects. In addition, in the process of extracting the partial discharge data through the MIMS program, the magnitude of amplitude is divided into the maximum value and the average value, and two pattern characteristics are set and compared and analyzed. In the first half of the proposed partial discharge pattern classifier, the input and hidden layers are classified by using the K-means clustering method and the output of the hidden layer is obtained. In the latter part, the cross entropy error function is used for parameter learning between the hidden layer and the output layer. The final output layer is output as a normalized probability value between 0 and 1 using the softmax function. The advantage of using the softmax function is that it allows access and application of multiple class problems and stochastic interpretation. First of all, there is an advantage that one output value affects the remaining output value and its accompanying learning is accelerated. Also, to solve the overfitting problem, L2-normalization is applied. To prove the superiority of the proposed pattern classifier, we compare and analyze the classification rate with conventional radial basis function neural networks.

• International Journal of Advanced Culture Technology
• /
• v.5 no.4
• /
• pp.57-62
• /
• 2017

In this paper, we applied long short term memory(LSTM) for classifying political polarity in cyber public sphere. The data collected from the cyber public sphere is transformed into word corpus data through word embedding. Based on this word corpus data, we train recurrent neural network (RNN) which is connected by LSTM's. Softmax function is applied at the output of the RNN. We conducted our proposed system to obtain experimental results, and we will enhance our proposed system by refining LSTM in our system.

• Nuclear Engineering and Technology
• /
• v.50 no.4
• /
• pp.582-588
• /
• 2018

Accident diagnosis is one of the complex tasks for nuclear power plant (NPP) operators. In abnormal or emergency situations, the diagnostic activity of the NPP states is burdensome though necessary. Numerous computer-based methods and operator support systems have been suggested to address this problem. Among them, the recurrent neural network (RNN) has performed well at analyzing time series data. This study proposes an algorithm for accident diagnosis using long short-term memory (LSTM), which is a kind of RNN, which improves the limitation for time reflection. The algorithm consists of preprocessing, the LSTM network, and postprocessing. In the LSTM-based algorithm, preprocessed input variables are calculated to output the accident diagnosis results. The outputs are also postprocessed using softmax to determine the ranking of accident diagnosis results with probabilities. This algorithm was trained using a compact nuclear simulator for several accidents: a loss of coolant accident, a steam generator tube rupture, and a main steam line break. The trained algorithm was also tested to demonstrate the feasibility of diagnosing NPP accidents.

• The Journal of Korean Institute of Information Technology
• /
• v.16 no.11
• /
• pp.105-113
• /
• 2018

In this paper, we proposed a method for effective classification of eye, nose, and mouth of human face. Most recent image classification uses Convolutional Neural Network(CNN). However, the features extracted by CNN are not sufficient and the classification effect is not too high. We proposed a new algorithm to improve the classification effect. The proposed method can be roughly divided into three parts. First, the Haar feature extraction algorithm is used to construct the eye, nose, and mouth dataset of face. The second, the model extracts CNN features of image using AlexNet. Finally, Haar-CNN features are extracted by performing convolution after Haar feature extraction. After that, CNN features and Haar-CNN features are fused and classify images using softmax. Recognition rate using mixed features could be increased about 4% than CNN feature. Experiments have demonstrated the performance of the proposed algorithm.

• Smart Structures and Systems
• /
• v.29 no.6
• /
• pp.757-766
• /
• 2022

To train deep learning algorithms, a sufficient number of data are required. However, in most engineering systems, the acquisition of fault data is difficult or sometimes not feasible, while normal data are secured. The dearth of data is one of the major challenges to developing deep learning models, and fault diagnosis in particular cannot be made in the absence of fault data. With this context, this paper proposes an anomaly detection methodology for rotating machines using only normal data with self-labeling. Since only normal data are used for anomaly detection, a self-labeling method is used to generate a new labeled dataset. The overall procedure includes the following three steps: (1) transformation of normal data to self-labeled data based on a pretext task, (2) training the convolutional neural networks (CNN), and (3) anomaly detection using defined anomaly score based on the softmax output of the trained CNN. The softmax value of the abnormal sample shows different behavior from the normal softmax values. To verify the proposed method, four case studies were conducted, on the Case Western Reserve University (CWRU) bearing dataset, IEEE PHM 2012 data challenge dataset, PHMAP 2021 data challenge dataset, and laboratory bearing testbed; and the results were compared to those of existing machine learning and deep learning methods. The results showed that the proposed algorithm could detect faults in the bearing testbed and compressor with over 99.7% accuracy. In particular, it was possible to detect not only bearing faults but also structural faults such as unbalance and belt looseness with very high accuracy. Compared with the existing GAN, the autoencoder-based anomaly detection algorithm, the proposed method showed high anomaly detection performance.