• International Journal of Internet, Broadcasting and Communication
• /
• v.14 no.1
• /
• pp.136-141
• /
• 2022

A very complex task in deep learning such as image classification must be solved with the help of neural networks and activation functions. The backpropagation algorithm advances backward from the output layer towards the input layer, the gradients often get smaller and smaller and approach zero which eventually leaves the weights of the initial or lower layers nearly unchanged, as a result, the gradient descent never converges to the optimum. We propose a two-factor non-saturating activation functions known as Bea-Mish for machine learning applications in deep neural networks. Our method uses two factors, beta (𝛽) and alpha (𝛼), to normalize the area below the boundary in the Mish activation function and we regard these elements as Bea. Bea-Mish provide a clear understanding of the behaviors and conditions governing this regularization term can lead to a more principled approach for constructing better performing activation functions. We evaluate Bea-Mish results against Mish and Swish activation functions in various models and data sets. Empirical results show that our approach (Bea-Mish) outperforms native Mish using SqueezeNet backbone with an average precision (AP50val) of 2.51% in CIFAR-10 and top-1accuracy in ResNet-50 on ImageNet-1k. shows an improvement of 1.20%.

• KIPS Transactions on Software and Data Engineering
• /
• v.9 no.10
• /
• pp.303-308
• /
• 2020

The ReLU(Rectified Linear Unit) function has been dominantly used as a standard activation function in most deep artificial neural network models since it was proposed. Later, Leaky ReLU, Swish, and Mish activation functions were presented to replace ReLU, which showed improved performance over existing ReLU function in image classification task. Therefore, we recognized the need to experiment with whether performance improvements could be achieved by replacing the RELU with other activation functions in the super resolution task. In this paper, the performance was compared by changing the activation functions in EDSR model, which showed stable performance in the super resolution task. As a result, in experiments conducted with changing the activation function of EDSR, when the resolution was converted to double, the existing activation function, ReLU, showed similar or higher performance than the other activation functions used in the experiment. When the resolution was converted to four times, Leaky ReLU and Swish function showed slightly improved performance over ReLU. PSNR and SSIM, which can quantitatively evaluate the quality of images, were able to identify average performance improvements of 0.06%, 0.05% when using Leaky ReLU, and average performance improvements of 0.06% and 0.03% when using Swish. When the resolution is converted to eight times, the Mish function shows a slight average performance improvement over the ReLU. Using Mish, PSNR and SSIM were able to identify an average of 0.06% and 0.02% performance improvement over the RELU. In conclusion, Leaky ReLU and Swish showed improved performance compared to ReLU for super resolution that converts resolution four times and Mish showed improved performance compared to ReLU for super resolution that converts resolution eight times. In future study, we should conduct comparative experiments to replace activation functions with Leaky ReLU, Swish and Mish to improve performance in other super resolution models.

• Journal of IKEEE
• /
• v.24 no.4
• /
• pp.1141-1147
• /
• 2020

The hornet species are so similar in shape that they are difficult for non-experts to classify, and because the size of the objects is small and move fast, it is more difficult to detect and classify the species in real time. In this paper, we developed a system that classifies hornets species in real time based on a deep learning algorithm using a boundary box. In order to minimize the background area included in the bounding box when labeling the training image, we propose a method of selecting only the head and body of the hornet. It also experimentally compares existing boundary box-based object recognition algorithms to find the best algorithms that can detect wasps in real time and classify their species. As a result of the experiment, when the mish function was applied as the activation function of the convolution layer and the hornet images were tested using the YOLOv4 model with the Spatial Attention Module (SAM) applied before the object detection block, the average precision was 97.89% and the average recall was 98.69%.

• Journal of the Korean Society of Radiology
• /
• v.18 no.1
• /
• pp.21-26
• /
• 2024

The one of the famous deep learning models for object detection task is you only look once version 5 (YOLOv5) framework based on the one stage architecture. In addition, YOLOv5 model indicated high performance for accurate lesion detection using the bottleneck CSP layer and skip connection function. The purpose of this study was to evaluate the performance of YOLOv5 framework according to various hyperparameters in position emission tomogrpahy (PET) phantom images. The dataset was obtained from QIN PET segmentation challenge in 500 slices. We set the bounding box to generate ground truth dataset using labelImg software. The hyperparameters for network train were applied by changing optimization function (SDG, Adam, and AdamW), activation function (SiLU, LeakyRelu, Mish, and Hardwish), and YOLOv5 model size (nano, small, large, and xlarge). The intersection over union (IOU) method was used for performance evaluation. As a results, the condition of outstanding performance is to apply AdamW, Hardwish, and nano size for optimization function, activation function and model version, respectively. In conclusion, we confirmed the usefulness of YOLOv5 network for object detection performance in nuclear medicine images.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.21 no.6
• /
• pp.117-122
• /
• 2021

Autonomous driving allows cars to drive without people and is being studied very actively thanks to the recent development of artificial intelligence technology. Among artificial intelligence technologies, deep reinforcement learning is used most effectively. Deep reinforcement learning requires us to build a neural network using an appropriate activation function. So far, many activation functions have been suggested, but different performances have been shown depending on the field of application. This paper compares and evaluates the performance of which activation function is effective when using deep reinforcement learning to learn autonomous driving on highways. To this end, the performance metrics to be used in the evaluation were defined and the values of the metrics according to each activation function were compared in graphs. As a result, when Mish was used, the reward was higher on average than other activation functions, and the difference from the activation function with the lowest reward was 9.8%.

• International Journal of Internet, Broadcasting and Communication
• /
• v.16 no.1
• /
• pp.63-71
• /
• 2024

Deep reinforcement learning (RL) significantly influences autonomous vehicle development by optimizing decision-making and adaptation to complex driving environments through simulation-based training. In deep RL, an activation function is used, and various activation functions have been proposed, but their performance varies greatly depending on the application environment. Therefore, finding the optimal activation function according to the environment is important for effective learning. In this paper, we analyzed nine commonly used activation functions for RL to compare and evaluate which activation function is most effective when using deep RL for autonomous vehicles to learn highway merging. To do this, we built a performance evaluation environment and compared the average reward of each activation function. The results showed that the highest reward was achieved using Mish, and the lowest using SELU. The difference in reward between the two activation functions was 10.3%.

• Journal of the Korea Society of Computer and Information
• /
• v.26 no.7
• /
• pp.29-36
• /
• 2021

In this paper, we classify ECG signal data for mobile devices using deep learning models. To classify abnormal heartbeats with high accuracy, three factors of the deep learning model are selected, and the classification accuracy is compared according to the changes in the conditions of the factors. We apply a CNN model that can self-extract features of ECG data and compare the performance of a total of 48 combinations by combining conditions of the depth of model, optimization method, and activation functions that compose the model. Deriving the combination of conditions with the highest accuracy, we obtained the highest classification accuracy of 97.88% when we applied 19 convolutional layers, an optimization method SGD, and an activation function Mish. In this experiment, we confirmed the suitability of feature extraction and abnormal beat detection of 1-channel ECG signals using CNN.

• Journal of Animal Science and Technology
• /
• v.65 no.3
• /
• pp.627-637
• /
• 2023

As the population and income levels rise, meat consumption steadily increases annually. However, the number of farms and farmers producing meat decrease during the same period, reducing meat sufficiency. Information and Communications Technology (ICT) has begun to be applied to reduce labor and production costs of livestock farms and improve productivity. This technology can be used for rapid pregnancy diagnosis of sows; the location and size of the gestation sacs of sows are directly related to the productivity of the farm. In this study, a system proposes to determine the number of gestation sacs of sows from ultrasound images. The system used the YOLOv7-E6E model, changing the activation function from sigmoid-weighted linear unit (SiLU) to a multi-activation function (SiLU + Mish). Also, the upsampling method was modified from nearest to bicubic to improve performance. The model trained with the original model using the original data achieved mean average precision of 86.3%. When the proposed multi-activation function, upsampling, and AutoAugment were applied, the performance improved by 0.3%, 0.9%, and 0.9%, respectively. When all three proposed methods were simultaneously applied, a significant performance improvement of 3.5% to 89.8% was achieved.

• Journal of the Korean Society of Radiology
• /
• v.16 no.1
• /
• pp.25-34
• /
• 2022

Brain computed tomography (CT) is useful for brain lesion diagnosis, such as brain hemorrhage, due to non-invasive methodology, 3-dimensional image provision, low radiation dose. However, there has been numerous misdiagnosis owing to a lack of radiologist and heavy workload. Recently, object detection technologies based on artificial intelligence have been developed in order to overcome the limitations of traditional diagnosis. In this study, the applicability of a deep learning-based YOLOv5s model was evaluated for brain hemorrhage detection using brain CT images. Also, the effect of hyperparameters in the trained YOLOv5s model was analyzed. The YOLOv5s model consisted of backbone, neck and output modules. The trained model was able to detect a region of brain hemorrhage and provide the information of the region. The YOLOv5s model was trained with various activation functions, optimizer functions, loss functions and epochs, and the performance of the trained model was evaluated in terms of brain hemorrhage detection accuracy and training time. The results showed that the trained YOLOv5s model is able to provide a bounding box for a region of brain hemorrhage and the accuracy of the corresponding box. The performance of the YOLOv5s model was improved by using the mish activation function, the stochastic gradient descent (SGD) optimizer function and the completed intersection over union (CIoU) loss function. Also, the accuracy and training time of the YOLOv5s model increased with the number of epochs. Therefore, the YOLOv5s model is suitable for brain hemorrhage detection using brain CT images, and the performance of the model can be maximized by using appropriate hyperparameters.