• Journal of information and communication convergence engineering
• /
• v.1 no.3
• /
• pp.157-162
• /
• 2003

We propose a learning method for improving the performance of the backpropagation algorithm. The proposed method is using a fuzzy logic system for automatic tuning of the learning rate of each weight. Instead of choosing a fixed learning rate, the fuzzy logic system is used to dynamically adjust the learning rate. The inputs of fuzzy logic system are delta and delta bar, and the output of fuzzy logic system is the learning rate. In order to verify the effectiveness of the proposed method, we performed simulations on the XOR problem, character classification, and function approximation. The results show that the proposed method considerably improves the performance compared to the general backpropagation, the backpropagation with momentum, and the Jacobs'delta-bar-delta algorithm.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.39 no.4
• /
• pp.19-27
• /
• 2002

We proposed an auto-tuning method of learning rate for performance improvement of backpropagation algorithm. Proposed method is used a fuzzy logic system for automatic tuning of learning rate. Instead of choosing a fixed learning rate, the fuzzy logic system is used to dynamically adjust learning rate. The inputs of fuzzy logic system are ${\Delta}$ and $\bar{{\Delta}}$, and the output is the learning rate. In order to verify the effectiveness of the proposed method, we performed simulations on a N-parity problem, function approximation, and Arabic numerals classification. The results show that the proposed method has considerably improved the performance compared to the backpropagation, the backpropagation with momentum, and the Jacobs' delta-bar-delta.

• Phonetics and Speech Sciences
• /
• v.15 no.3
• /
• pp.75-82
• /
• 2023

Automatic speech recognition (ASR) has been revolutionized with deep learning-based approaches, among which self-supervised learning methods have proven to be particularly effective. In this study, we aim to enhance the performance of OpenAI's Whisper model, a multilingual ASR system on the Korean language. Whisper was pretrained on a large corpus (around 680,000 hours) of web speech data and has demonstrated strong recognition performance for major languages. However, it faces challenges in recognizing languages such as Korean, which is not major language while training. We address this issue by fine-tuning the Whisper model with an additional dataset comprising about 1,000 hours of Korean speech. We also compare its performance against a Transformer model that was trained from scratch using the same dataset. Our results indicate that fine-tuning the Whisper model significantly improved its Korean speech recognition capabilities in terms of character error rate (CER). Specifically, the performance improved with increasing model size. However, the Whisper model's performance on English deteriorated post fine-tuning, emphasizing the need for further research to develop robust multilingual models. Our study demonstrates the potential of utilizing a fine-tuned Whisper model for Korean ASR applications. Future work will focus on multilingual recognition and optimization for real-time inference.

• Journal of Multimedia Information System
• /
• v.6 no.4
• /
• pp.209-216
• /
• 2019

Over the past decade, researchers were able to solve complex medical problems as well as acquire deeper understanding of entire issue due to the availability of machine learning techniques, particularly predictive algorithms and automatic recognition of patterns in medical imaging. In this study, a technique called transfer learning has been utilized to classify Magnetic Resonance (MR) images by a pre-trained Convolutional Neural Network (CNN). Rather than training an entire model from scratch, transfer learning approach uses the CNN model by fine-tuning them, to classify MR images into Alzheimer's disease (AD), mild cognitive impairment (MCI) and normal control (NC). The performance of this method has been evaluated over Alzheimer's Disease Neuroimaging (ADNI) dataset by changing the learning rate of the model. Moreover, in this study, in order to demonstrate the transfer learning approach we utilize different pre-trained deep learning models such as GoogLeNet, VGG-16, AlexNet and ResNet-18, and compare their efficiency to classify AD. The overall classification accuracy resulted by GoogLeNet for training and testing was 99.84% and 98.25% respectively, which was exceptionally more than other models training and testing accuracies.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.40 no.1
• /
• pp.10-18
• /
• 2003

In this paper, we proposed a learning algorithm for the ADALINE network. The proposed algorithm exploits fuzzy system for automatic tuning of the weight parameters of the ADALINE network. The inputs of the fuzzy system are error and change of error, and the output is the weight variation. We used different scaling factor for each weights. In order to verify the effectiveness of the proposed algorithm, we peformed the simulation and experimentation for the cases of the noise cancellation and the inverted pendulum control. The results show that the proposed algorithm does not need the learning rate and improves 4he performance compared to the Widrow-Hoff delta rule for ADALINE.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.38 no.6
• /
• pp.26-37
• /
• 2001

In order to improve the several problems of the general backpropagation, we propose a method using a fuzzy logic system for automatic tuning of the activation function gain in the backpropagation. First, we researched that the changing of the gain of sigmoid function is equivalent to changing the learning rate, the weights, and the biases. The inputs of the fuzzy logic system were the sensitivity of error respect to the last layer and the mean sensitivity of error respect to the hidden layer, and the output was the gain of the sigmoid function. In order to verify the effectiveness of the proposed method, we performed simulations on the parity problem, function approximation, and pattern recognition. The results show that the proposed method has considerably improved the performance compared to the general backpropagation.