Minho Ha's research while affiliated with Pohang University of Science and Technology and other places
What is this page?
This page lists the scientific contributions of an author, who either does not have a ResearchGate profile, or has not yet added these contributions to their profile.
It was automatically created by ResearchGate to create a record of this author's body of work. We create such pages to advance our goal of creating and maintaining the most comprehensive scientific repository possible. In doing so, we process publicly available (personal) data relating to the author as a member of the scientific community.
If you're a ResearchGate member, you can follow this page to keep up with this author's work.
If you are this author, and you don't want us to display this page anymore, please let us know.
It was automatically created by ResearchGate to create a record of this author's body of work. We create such pages to advance our goal of creating and maintaining the most comprehensive scientific repository possible. In doing so, we process publicly available (personal) data relating to the author as a member of the scientific community.
If you're a ResearchGate member, you can follow this page to keep up with this author's work.
If you are this author, and you don't want us to display this page anymore, please let us know.
Publications (7)
In order to effectively reduce buffer energy consumption, which constitutes a significant part of the total energy consumption in a convolutional neural network (CNN), it is useful to apply different amounts of energy conservation effort to the different levels of a CNN as the buffer energy to total energy usage ratios can differ quite substantiall...
![FIGURE 1. The concept of DCS-CNN [14] targeting (a) Gaussian noise...](publication/335668776/figure/fig1/AS:800302005510144@1567818275389/The-concept-of-DCS-CNN-14-targeting-a-Gaussian-noise-level-2-case-and-b-Gaussian_Q320.jpg)
![FIGURE 2. MixQualNet architecture proposed in [12].](publication/335668776/figure/fig2/AS:800302009708544@1567818276490/MixQualNet-architecture-proposed-in-12_Q320.jpg)
![FIGURE 5. Example CIFAR-100 image [17] after various level of Gaussian...](publication/335668776/figure/fig3/AS:800302030663680@1567818281746/Example-CIFAR-100-image-17-after-various-level-of-Gaussian-noise-Gaussian-blur-and_Q320.jpg)
![FIGURE 6. Example Caltech-256 image [18] after various level of...](publication/335668776/figure/fig4/AS:800302034853888@1567818282260/Example-Caltech-256-image-18-after-various-level-of-Gaussian-noise-Gaussian-blur-and_Q320.jpg)
Neural networks trained using images with a certain type of distortion should be better at classifying test images with the same type of distortion than generally-trained neural networks, given other factors being equal. Based on this observation, an ensemble of convolutional neural networks (CNNs) trained with different types and degrees of distor...
Deep convolutional neural networks have proven to be well suited for image classification applications. However, if there is distortion in the image, the classification accuracy can be significantly degraded, even with state-of-the-art neural networks. The accuracy cannot be significantly improved by simply training with distorted images. Instead,...
Approximate multiplication is a common operation used in approximate computing methods for high performance and low power computing. Power-efficient circuits for approximate multiplication can be realized with an approximate 4-2 compressor. This letter presents a novel design that uses a modification of a previous approximate 4-2 compressor design...
This brief presents a hardware-efficient logarithm circuit design based on a novel discontinuous piecewise linear approximation method. Hardware synthesis results targeted for a commercial application specific integrated circuit cell library and field-programmable gate array show the practicality of the proposed design. A new figure of merit that c...
Citations
... Paper [76] presents an approach to voltage control of CNN structural elements to reduce energy by computational errors moderately increasing. It is based on layer-by-layer scaling of the buffer voltage based on the error tolerance analysis. ...
... Work [75] proposes a selective neural network system for distorted image classifying. This system is an ensemble of CNNs, each of which is designed to process distorted image by a certain noise type. ...
... Most edge devices do not process the training process that requires processing large amounts of data and only proceed with inference. Nevertheless, the CNN layer becomes gradually deeper, and studies to reduce the amount of memory access and computation, such as pruning, quantization, and compression, have also been actively conducted [11]- [14]. ...
... Then, by encoding the inputs using the generate and propagate signals, its error is significantly reduced. In [43], an approximate 4:2 compressor and an error recovery module are employed to reduce error in results. Another work [44] introduces an approximate 4:2 compressor and a modified architecture for Dadda multiplier. ...
... Autonomous cars, cloud computing and image recognition commonly employ 5G technology, because these applications need complex arithmetic operations such as multiplication, exponentiation and division, which require complex implementation area and longer processing time. To address these issues, logarithmic conversion systems [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20] have been discussed over the recent decades. Logarithmic conversion systems comprise logarithmic converters, simple arithmetic units and antilogarithmic converters. ...
