• KIPS Transactions on Software and Data Engineering
• /
• v.7 no.7
• /
• pp.275-280
• /
• 2018

As the global population becomes aging, the demand for health services for the elderly is expected to increase. In particular, The elderly visiting the emergency department sometimes have complex medical, social, and physical problems, such as having a variety of illnesses or complaints of unusual symptoms. The proposed system is designed to predict the mortality of the elderly patients who are over 65 years old and have admitted the emergency department. For mortality prediction, we compare the support vector machines and Feed Forward Neural Network (FFNN) trained with medical data such as age, sex, blood pressure, body temperature, etc. The results of the FFNN with a hidden layer are best in the mortality prediction, and F1 score and the AUC is 52.0%, 88.6% respectively. If we improve the performance of the proposed system by extracting better medical features, we will be able to provide better medical services through an effective and quick allocation of medical resources for the elderly patients visiting the emergency department.

• Physical Therapy Korea
• /
• v.14 no.2
• /
• pp.21-28
• /
• 2007

The purpose of this study was to compare the onset times of muscle activities and the order of muscle firing in erector spinae, multifidus, rectus abdominis and biceps brachii during perturbation between subjects with and without work-related chronic low back pain (LBP). Twenty-nine subjects, 14 with and 15 without LBP, participated in this study. The muscle responses were measured by surface EMG (electromyography) during perturbation in eye opened and eye closed conditions. The EMG onset times of the erector spinae, multifidus, rectus abdominis and biceps brachii were similar between groups in eye closed condition. But the onset times of the erector spinae, multifidus, rectus abdominis were significantly delayed in subjects with LBP in eye opened condition. The results provide an evidence for impaired feed-forward control of the trunk muscles in subjects with LBP. Further studies are needed to identify whether the impaired feed-forward control of the trunk muscles is the contributing factor to LBP.

• Journal of Power Electronics
• /
• v.14 no.5
• /
• pp.1047-1056
• /
• 2014

In most grid-connected inverters (GCI) with an LCL filter, since the design of both the LCL filter and the controller is done separately, considerable tuning efforts have to be exerted when compared to inverters using an L filter. Consequently, an integrated co-design of the filter and the controller for an LCL-type GCI is proposed in this paper. The control strategy includes only a PI current controller and a proportional grid voltage feed-forward controller. The capacitor is removed from the LCL filer and the design procedure starts from an L-type GCI with a PI current controller. After the PI controller has been settled, the capacitor is added back to the filter. Hence, it introduces a resonance frequency, which is identified based on the crossover frequencies to accommodate the preset PI controller. Using the proposed co-design method, harmonic standards are satisfied and other practical constraints are met. Furthermore, the grid voltage feed-forward control can bring an inherent damping characteristic. In such a way, the good control performance offered by the original L-type GCI and the sharp harmonic attenuation offered by the latter designed LCL filter can be well integrated. Moreover, only the grid current and grid voltage are sensed. Simulation and experimental results verify the feasibility of the proposed design methodology.

• Journal of IKEEE
• /
• v.21 no.4
• /
• pp.331-337
• /
• 2017

In this paper, we present the design of a 20MHz bandwidth 3rd-order continuous-time low-pass sigma-delta modulator with low-noise and low-power consumption. The bandwidth of the system is sufficient to accommodate LTE and other wireless network standards. The 3rd-order low-pass filter with feed-forward architecture achieves the low-power consumption as well as the low complexity. The system uses 3bit flash quantizer to provide fast data conversion. The current-steering DAC achieves low-power and improved sensitivity without additional circuitries. Cross-coupled transistors are adopted to reduce the current glitches. The proposed system achieves a peak SNDR of 65.9dB with 20MHz bandwidth and power consumption of 32.65mW. The in-band IM3 is simulated to be 69dBc with 600mVp-p two tone input tones. The circuit is designed in a 0.18-um CMOS technology and is driven by 500MHz sampling rate signal.

• International Journal of Naval Architecture and Ocean Engineering
• /
• v.13 no.1
• /
• pp.641-649
• /
• 2021

Fuel oil consumption (FOC) must be minimized to determine the economic route of a ship; hence, the ship power must be predicted prior to route planning. For this purpose, a numerical method using test results of a model has been widely used. However, predicting ship power using this method is challenging owing to the uncertainty of the model test. An onboard test should be conducted to solve this problem; however, it requires considerable resources and time. Therefore, in this study, a deep feed-forward neural network (DFN) is used to predict ship power using deep learning methods that involve data pattern recognition. To use data in the DFN, the input data and a label (output of prediction) should be configured. In this study, the input data are configured using ocean environmental data (wave height, wave period, wave direction, wind speed, wind direction, and sea surface temperature) and the ship's operational data (draft, speed, and heading). The ship power is selected as the label. In addition, various treatments have been used to improve the prediction accuracy. First, ocean environmental data related to wind and waves are preprocessed using values relative to the ship's velocity. Second, the structure of the DFN is changed based on the characteristics of the input data. Third, the prediction accuracy is analyzed using a combination comprising five hyperparameters (number of hidden layers, number of hidden nodes, learning rate, dropout, and gradient optimizer). Finally, k-means clustering is performed to analyze the effect of the sea state and ship operational status by categorizing it into several models. The performances of various prediction models are compared and analyzed using the DFN in this study.

• Journal of IKEEE
• /
• v.28 no.1
• /
• pp.53-64
• /
• 2024

With the advancement of NLP(Natural Language Processing) models, conversational AI such as ChatGPT is becoming increasingly popular. To enhance processing speed and reduce power consumption, it is important to implement the Transformer algorithm, which forms the basis of the latest natural language processing models, in hardware. In particular, the multi-head attention and feed-forward network, which analyze the relationships between different words in a sentence through matrix multiplication, are the most computationally intensive core algorithms in the Transformer. In this paper, we propose a new variable systolic array based on the number of input words to enhance matrix multiplication speed. Quantization maintains Transformer accuracy, boosting memory efficiency and speed. For evaluation purposes, this paper verifies the clock cycles required in multi-head attention and feed-forward network and compares the performance with other multipliers.

• Structural Engineering and Mechanics
• /
• v.46 no.5
• /
• pp.673-693
• /
• 2013

This paper presents a systematic design and training procedure for the feed-forward back-propagation neural network (NN) modeling of both forward and inverse behavior of a rotary magnetorheological (MR) damper based on experimental data. For the forward damper model, with damper force as output, an optimization procedure demonstrates accurate training of the NN architecture with only current and velocity as input states. For the inverse damper model, with current as output, the absolute value of velocity and force are used as input states to avoid negative current spikes when tracking a desired damper force. The forward and inverse damper models are trained and validated experimentally, combining a limited number of harmonic displacement records, and constant and half-sinusoidal current records. In general the validation shows accurate results for both forward and inverse damper models, where the observed modeling errors for the inverse model can be related to knocking effects in the measured force due to the bearing plays between hydraulic piston and MR damper rod. Finally, the validated models are used to emulate pure viscous damping. Comparison of numerical and experimental results demonstrates good agreement in the post-yield region of the MR damper, while the main error of the inverse NN occurs in the pre-yield region where the inverse NN overestimates the current to track the desired viscous force.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.51 no.10
• /
• pp.57-63
• /
• 2014

This paper proposes an architecture of $3^{rd}$ order SDM(Sigma-Delta Modulator) with delayed feed-forward path in order to reduce the power consumption and area. The proposed SDM improve the architecture of conventional $3^{rd}$ order SDM which consists of two integrators. The proposed architecture can increase the coefficient values of first stage doubly by inserting the delayed feed-forward path. Accordingly, compared with the conventional architecture, the capacitor value($C_I$) of first integrator is reduced by half. Thus, because the load capacitance of first integrator became the half of original value, the output current of first op-amp is reduced as 51% and the capacitance area of first integrator is reduced as 48%. Therefore, the proposed method can optimize the power and the area. The proposed architecture in this paper is simulated under conditions which are supply voltage of 1.8V, input signal 1Vpp/1KHz, signal bandwidth of 24KHz and sampling frequency of 2.8224MHz in the 0.18um CMOS process. The simulation results are SNR(Signal to Noise Ratio) of 88.9dB and ENOB(Effective Number of Bits) of 14-bits. The total power consumption of the proposed SDM is $180{\mu}W$.

• Journal of Internet Computing and Services
• /
• v.20 no.5
• /
• pp.95-104
• /
• 2019

Various kinds of devices are used for the Internet of Things (IoT) service, and IoT devices mainly use communication technology that uses the frequency of the unlicensed band. There are several types of communication technology in the unlicensed band, but WiFi is most commonly used. Devices used for IoT services vary in computing resources from devices with limited capabilities to smartphones and provide services over wireless networks such as WiFi. Most IoT devices can't perform complex operations for network control, thus they choose a WiFi access point (AP) based on signal strength. This causes a decrease in IoT service efficiency. In this paper, an intelligent AP system that can efficiently control the WiFi connection of the IoT devices is implemented. Based on the network information measured by the IoT device, the access point learns using a feed forward neural network algorithm, and predicts a network connection state to control the WiFi connection. By controlling the WiFi connection at the AP, the service efficiency of the IoT device can be improved.

• Journal of information and communication convergence engineering
• /
• v.6 no.2
• /
• pp.154-157
• /
• 2008

A class-E RF(Radio Frequency) power amplifier for wireless application is designed using standard CMOS technology. To drive the class-E power amplifier, a class-F RF power amplifier is used and the reliability characteristics are studied with a class-E load network. The reliability characteristic is improved when a finite-DC feed inductor is used instead of an RF choke with the load. After one year of operating, when the load is an RF choke the output current and voltage of the power amplifier decrease about 17% compared to initial values. But when the load is a finite DC-feed inductor the output current and voltage decrease 9.7%. The S-parameter such as input reflection coefficient(S11) and the forward transmission scattering parameter(S21) is simulated with the stress time. In a finite DC-feed inductor the characteristics of S-parameter are changed slightly compared to an RF-choke inductor. From the simulation results, the class-E power amplifier with a finite DC-feed inductor shows superior reliability characteristics compared to power amplifier using an RF choke.