• The Journal of the Acoustical Society of Korea
• /
• v.22 no.3E
• /
• pp.124-132
• /
• 2003

We propose a new method for speech/non-speech classifiers based on concepts of the adaptive boosting (AdaBoost) algorithm in order to detect speech for robust speech recognition. The method uses a combination of simple base classifiers through the AdaBoost algorithm and a set of optimized speech features combined with spectral subtraction. The key benefits of this method are the simple implementation, low computational complexity and the avoidance of the over-fitting problem. We checked the validity of the method by comparing its performance with the speech/non-speech classifier used in a standard voice activity detector. For speech recognition purpose, additional performance improvements were achieved by the adoption of new features including speech band energies and MFCC-based spectral distortion. For the same false alarm rate, the method reduced 20-50% of miss errors.

• International Journal of Automotive Technology
• /
• v.8 no.5
• /
• pp.637-650
• /
• 2007

A new on-line fault detection and isolation(FDI) scheme has been proposed for engines using an adaptive neural network classifier; this paper investigates the robustness of this scheme by evaluating in a wide range of operational modes. The neural classifier is made adaptive to cope with the significant parameter uncertainty, disturbances, and environmental changes. The developed scheme is capable of diagnosing faults in the on-line mode and can be directly implemented in an on-board diagnosis system(hardware). The robustness of the FDI for the closed-loop system with crankshaft speed feedback is investigated by testing it for a wide range of operational modes, including robustness against fixed and sinusoidal throttle angle inputs, change in load, change in an engine parameter, and all changes occurring simultaneously. The evaluations are performed using a mean value engine model(MVEM), which is a widely used benchmark model for engine control system and FDI system design. The simulation results confirm the robustness of the proposed method for various uncertainties and disturbances.

• Journal of Digital Contents Society
• /
• v.19 no.1
• /
• pp.133-140
• /
• 2018

Accurate sentiment classification is an important research topic in sentiment analysis. This study suggests an efficient classification method of Korean sentiment using word2vec and ensemble methods which have been recently studied variously. For the 200,000 Korean movie review texts, we generate a POS-based BOW feature and a feature using word2vec, and integrated features of two feature representation. We used a single classifier of Logistic Regression, Decision Tree, Naive Bayes, and Support Vector Machine and an ensemble classifier of Adaptive Boost, Bagging, Gradient Boosting, and Random Forest for sentiment classification. As a result of this study, the integrated feature representation composed of BOW feature including adjective and adverb and word2vec feature showed the highest sentiment classification accuracy. Empirical results show that SVM, a single classifier, has the highest performance but ensemble classifiers show similar or slightly lower performance than the single classifier.

• Journal of the Korean Society for Nondestructive Testing
• /
• v.15 no.4
• /
• pp.520-530
• /
• 1996

In this study, the artificial defects in rotary compressor are classified using pattern recognition of acoustic emission signal. For this purpose the computer program is developed. The neural network classifier is compared with the statistical classifier such as the linear discriminant function classifier and empirical Bayesian classifier. It is concluded that the former is better. It is possible to acquire the recognition rate of above 99% by neural network classifier.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 1998.10a
• /
• pp.361-367
• /
• 1998

The main two applications of the Genetic Algorithms(GA) are the optimization and the machine learning. Machine Learning has two objectives that make the complex system learn its environment and produce the proper output of a system. The machine learning using the Genetic Algorithms is called GA machine learning or genetic-based machine learning (GBML). The machine learning is different from the optimization problems in finding the rule set. In optimization problems, the population of GA should converge into the best individual because optimization problems, the population of GA should converge into the best individual because their objective is the production of the individual near the optimal solution. On the contrary, the machine learning systems need to find the set of cooperative rules. There are two methods in GBML, Michigan method and Pittsburgh method. The former is that each rule is expressed with a string, the latter is that the set of rules is coded into a string. Th classifier system of Holland is the representative model of the Michigan method. The classifier systems arrange the strength of classifiers of classifier list using the message list. In this method, the real time process and on-line learning is possible because a set of rule is adjusted on-line. A classifier system has three major components: Performance system, apportionment of credit system, rule discovery system. In this paper, we solve the food search problem with the learning and evolution of an artificial ant using the learning classifier system.

• 제어로봇시스템학회:학술대회논문집
• /
• 2000.10a
• /
• pp.367-367
• /
• 2000

In this paper, an Adaptive neuro-fuzzy Inference system(ANFIS) using fuzzy min-max network(FMMN) is proposed. Fuzzy min-max network classifier that utilizes fuzzy sets as pattern classes is described. Each fuzzy set is an aggregation of fuzzy set hyperboxes. Here, the proposed method transforms the hyperboxes into gaussian membership functions, where the transformed membership functions are inserted for generating fuzzy rules of ANFIS. Finally, we applied the proposed method to the classification problem of iris data and obtained a better performance than previous works.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.52 no.8
• /
• pp.97-105
• /
• 2015

Content-adaptive steganography methods embed secret messages in hard-to-model regions of covers such as complicated texture or noisy area. Content-adaptive steganalysis methods often need high dimensional features to capture more subtle relationships of local dependencies among adjacent pixels. However, these methods require many computational complexity and depend on the location of hidden message and the exploited distortion metrics. In this paper, we propose an improved steganalysis method for content-adaptive steganography to enhance detection rate with small number features. We first show that the features form the difference between DCT coefficients are useful for analyzing the content-adaptive steganography methods, and present feature extraction mehtod using first-order Markov probability for the the difference between DCT coefficients. The extracted features are used as input of ensemble classifier. Experimental results show that the proposed method outperforms previous schemes in terms of detection rates and accuracy in spite of a small number features in various content-adaptive stego images.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.13 no.10
• /
• pp.4865-4885
• /
• 2019

We propose a FPGA based design that performs real-time power-efficient analysis of heterogeneous sensor data using adaptive ANN on edge gateway of smart military wearables. In this work, four independent ANN classifiers are developed with optimum topologies. Out of which human activity, BP and toxic gas classifier are multiclass and ECG classifier is binary. These classifiers are later integrated into a single adaptive ANN hardware with a select line(s) that switches the hardware architecture as per the sensor type. Five versions of adaptive ANN with different precisions have been synthesized into IP cores. These IP cores are implemented and tested on Xilinx Artix-7 FPGA using Microblaze test system and LabVIEW based sensor simulators. The hardware analysis shows that the adaptive ANN even with 8-bit precision is the most efficient IP core in terms of hardware resource utilization and power consumption without compromising much on classification accuracy. This IP core requires only 31 microseconds for classification by consuming only 12 milliwatts of power. The proposed adaptive ANN design saves 61% to 97% of different FPGA resources and 44% of power as compared with the independent implementations. In addition, 96.87% to 98.75% of data throughput reduction is achieved by this edge gateway.

• Proceedings of the KIEE Conference
• /
• 2004.11c
• /
• pp.162-164
• /
• 2004

This paper proposes a new classification system combining the adaptive feature weighting algorithm using the genetic algorithm and the modified KNN rule. GA is employed to choose the middle value of weights and weights of features for high performance of the system. The modified KNN rule is proposed to estimate the class of test pattern using adaptive feature space. Experiments with the unconstrained handwritten digit database of Concordia University in Canada are conducted to show the performance of the proposed method.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.9 no.11
• /
• pp.4573-4584
• /
• 2015

The traditional feature extraction algorithms for recognition of communication signals can hardly realize the balance between computational complexity and signals' interclass gathered degrees. They can hardly achieve high recognition rate at low SNR conditions. To solve this problem, a novel feature extraction algorithm based on Holder coefficient was proposed, which has the advantages of low computational complexity and good interclass gathered degree even at low SNR conditions. In this research, the selection methods of parameters and distribution properties of the extracted features regarding Holder coefficient theory were firstly explored, and then interval gray relation algorithm with improved adaptive weight was adopted to verify the effectiveness of the extracted features. Compared with traditional algorithms, the proposed algorithm can more accurately recognize signals at low SNR conditions. Simulation results show that Holder coefficient based features are stable and have good interclass gathered degree, and interval gray relation classifier with adaptive weight can achieve the recognition rate up to 87% even at the SNR of -5dB.