• International Journal of Advanced Culture Technology
• /
• v.10 no.2
• /
• pp.246-251
• /
• 2022

As the number of data and devices explodes, centralized data processing and AI analysis have limitations due to the load on the network and cloud. On-device AI technology can provide intelligent services without overloading the network and cloud because the device itself performs AI models. Accordingly, the need for on-device AI technology is emerging. Many smartphones are equipped with On-Device AI technology to support the use of related functions. In this paper, we compare software tools that implement On-Device AI.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.10a
• /
• pp.597-599
• /
• 2022

When IoT device performs on-device AI, the device is required to use various AI models selectively according to target service and surrounding environment. Also, AI model can be updated by additional training such as federated learning or adapting the improved technique. Hence, for successful on-device AI, IoT device should acquire various AI models selectively or update previous AI model to new one. In this paper, we propose AI model repository to tackle this issue. The repository supports AI model registration, searching, management, and deployment along with dashboard for practical usage. We implemented it using Node.js and Vue.js to verify it works well.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.22 no.5
• /
• pp.17-22
• /
• 2022

In this paper, we design a lightweight embedded device that can support intelligent edge computing, and show that the device quickly detects an object in an image input from a camera device in real time. The proposed system can be applied to environments without pre-installed infrastructure, such as an intelligent video control system for industrial sites or military areas, or video security systems mounted on autonomous vehicles such as drones. The On-Device AI(Artificial intelligence) technology is increasingly required for the widespread application of intelligent vision recognition systems. Computing offloading from an image data acquisition device to a nearby edge device enables fast service with less network and system resources than AI services performed in the cloud. In addition, it is expected to be safely applied to various industries as it can reduce the attack surface vulnerable to various hacking attacks and minimize the disclosure of sensitive data.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.24 no.1
• /
• pp.1-8
• /
• 2024

On-device AI technology, which can operate AI models at the edge devices to support real-time processing and privacy enhancement, is attracting attention. As intelligent IoT is applied to various industries, services utilizing the on-device AI technology are increasing significantly. However, general deep learning models require a lot of computational resources for inference and learning. Therefore, various lightweighting methods such as quantization and pruning have been suggested to operate deep learning models in embedded edge devices. Among the lightweighting methods, we analyze how to lightweight and apply deep learning models to edge computing devices, focusing on pruning technology in this paper. In particular, we utilize dynamic and static pruning techniques to evaluate the inference speed, accuracy, and memory usage of a lightweight AI vision model. The content analyzed in this paper can be used for intelligent video control systems or video security systems in autonomous vehicles, where real-time processing are highly required. In addition, it is expected that the content can be used more effectively in various IoT services and industries.

• Korean Journal of Radiology
• /
• v.22 no.3
• /
• pp.442-453
• /
• 2021

Artificial intelligence (AI) will likely affect various fields of medicine. This article aims to explain the fundamental principles of clinical validation, device approval, and insurance coverage decisions of AI algorithms for medical diagnosis and prediction. Discrimination accuracy of AI algorithms is often evaluated with the Dice similarity coefficient, sensitivity, specificity, and traditional or free-response receiver operating characteristic curves. Calibration accuracy should also be assessed, especially for algorithms that provide probabilities to users. As current AI algorithms have limited generalizability to real-world practice, clinical validation of AI should put it to proper external testing and assisting roles. External testing could adopt diagnostic case-control or diagnostic cohort designs. A diagnostic case-control study evaluates the technical validity/accuracy of AI while the latter tests the clinical validity/accuracy of AI in samples representing target patients in real-world clinical scenarios. Ultimate clinical validation of AI requires evaluations of its impact on patient outcomes, referred to as clinical utility, and for which randomized clinical trials are ideal. Device approval of AI is typically granted with proof of technical validity/accuracy and thus does not intend to directly indicate if AI is beneficial for patient care or if it improves patient outcomes. Neither can it categorically address the issue of limited generalizability of AI. After achieving device approval, it is up to medical professionals to determine if the approved AI algorithms are beneficial for real-world patient care. Insurance coverage decisions generally require a demonstration of clinical utility that the use of AI has improved patient outcomes.

• Journal of information and communication convergence engineering
• /
• v.20 no.3
• /
• pp.166-173
• /
• 2022

Recently, studies have been conducted on intelligent transportation systems (ITS) that provide safety and convenience to humans. Systems that compose the ITS adopt architectures that applied the cloud computing which consists of a high-performance general-purpose processor or graphics processing unit. However, an architecture that only used the cloud computing requires a high network bandwidth and consumes much power. Therefore, applying edge computing to ITS is essential for solving these problems. In this paper, we propose an edge artificial intelligence (AI) device based ITS. Edge AI which is applicable to various systems in ITS has been applied to license plate recognition. We implemented edge AI on a field-programmable gate array (FPGA). The accuracy of the edge AI for license plate recognition was 0.94. Finally, we synthesized the edge AI logic with Magnachip/Hynix 180nm CMOS technology and the power consumption measured using the Synopsys's design compiler tool was 482.583mW.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.19 no.3
• /
• pp.151-158
• /
• 2024

Although on-device artificial intelligence (AI) has gained attention to diagnosing machine faults in real time, most previous studies did not consider the model retraining and redeployment processes that must be performed in real-world industrial environments. Our study addresses this challenge by proposing an on-device AI-based real-time machine fault diagnosis system that utilizes continual learning. Our proposed system includes a lightweight convolutional neural network (CNN) model, a continual learning algorithm, and a real-time monitoring service. First, we developed a lightweight 1D CNN model to reduce the cost of model deployment and enable real-time inference on the target edge device with limited computing resources. We then compared the performance of five continual learning algorithms with three public bearing fault datasets and selected the most effective algorithm for our system. Finally, we implemented a real-time monitoring service using an open-source data visualization framework. In the performance comparison results between continual learning algorithms, we found that the replay-based algorithms outperformed the regularization-based algorithms, and the experience replay (ER) algorithm had the best diagnostic accuracy. We further tuned the number and length of data samples used for a memory buffer of the ER algorithm to maximize its performance. We confirmed that the performance of the ER algorithm becomes higher when a longer data length is used. Consequently, the proposed system showed an accuracy of 98.7%, while only 16.5% of the previous data was stored in memory buffer. Our lightweight CNN model was also able to diagnose a fault type of one data sample within 3.76 ms on the Raspberry Pi 4B device.

• Asian-Australasian Journal of Animal Sciences
• /
• v.16 no.9
• /
• pp.1268-1273
• /
• 2003

Sixty postpartum lactating Friesian cows in 3 treatments at a commercial dairy farm were used to study the effect of using progesterone supplementation with GnRH and PGF2$\alpha$ synchronization with and without timed AI on fertility during summer. Cows in treatment1($Tr_1$) and treatment2 ($Tr_1$) were fitted with progesterone releasing intravaginal device (PRID) device and injected with 10 g GnRH agonist on $51{\pm}3$ d postpartum (pp). Seven days later, PRID was removed and cows received 25 mg PGF2$\alpha$. Two days later, $Tr_1$ cows received another injection of 10 g GnRH and timed AI 16-20 h later. Control cows received only 25 mg PGF2$\alpha$ $58{\pm}3d\;pp$. $Tr_2$ and control cows were AI at detected estrus. Serum progesterone for all cows was determined on days of injection, AI and 21, 23 and 28 d postinsemination. Pregnancy rates from first AI based on serum P4 concentrations on d 21, 23 and 28 postinsemination (50, 40 and 35%) and that based on rectal palpation 40-45 d postinsemination (30, 15 and 15% for $Tr_1$, $Tr_2$ and control cows, respectively) did not differ among the three groups. Whereas, pregnancy rate at 120 d pp for $Tr_1$ (65%) was higher (p<0.05) than that in $Tr_2$ (30%) or control (30%). The overall pregnancy rate was not significantly different (90, 90 and 75% for $Tr_1$, $Tr_2$ and control, respectively). Days open for cows in $Tr_1$ ($100.3{\pm}9$) was less (p<0.03) than that in $Tr_2$ ($130.9{\pm}9$) or control ($135.1{\pm}10$). Results indicate that using PRID device with Ovsynch program had significantly increased pregnancy rate and decreased days open compared to AI at detected estrus after synchronization with GnRH, PRID and PGF2$\alpha$ or synchronization with one injection of PGF2$\alpha$.

• Journal of Biomedical Engineering Research
• /
• v.44 no.6
• /
• pp.428-442
• /
• 2023

In this study, we have a goal to develop usability evaluation guidelines for heuristic-based artificial intelligence-based Software as a Medical Device (SaMD) in the medical field. We conducted a gap analysis between medical hardware (H/W) and non-medical software (S/W) based on ten heuristic principles. Through severity assessments, we identified 69 evaluation domains and 112 evaluation criteria aligned with the ten heuristic principles. Subsequently, we categorized each evaluation domain into five types, including user safety, data integrity, regulatory compliance, patient therapeutic effectiveness, and user convenience. We proposed usability evaluation guidelines that apply the newly derived heuristic-based Software as a Medical Device (SaMD) evaluation factors to the risk management process. In the discussion, we also have proposed the potential applications of the research findings and directions for future research. We have emphasized the importance of the judicious application of AI technology in the medical field and the evaluation of usability evaluation and offered valuable guidelines for various stakeholders, including medical device manufacturers, healthcare professionals, and regulatory authorities.

• The Journal of Korean Medicine
• /
• v.30 no.3
• /
• pp.98-105
• /
• 2009

Objective: Increased aortic and carotid arterial augmentation index (AI) is associated with the risk of cardiovascular disease. The most widely used approach for determining central arterial AI is by calculating the aortic pressure waveform from radial arterial waveforms using a transfer function. But how the change of waveform by applied pressure and the pattern of the change rely on subject's characteristics has not been recognized. In this study, we use a new method for measuring radial waveform and observe the change of waveform and the deviation of radial AI in the same position by applied pressure. Method: Forty-six non-patient volunteers (31 men and 15 women, age range 21-58 years) were enrolled for this study. Informed consent in a form approved by the institutional review board was obtained in all subjects. Blood pressure was measured on the left upper arm using an oscillometric method, radial pressure waves were recorded with the use of an improved automated tonometry device. DMP-3000(DAEYOMEDI Co., Ltd. Ansan, Korea) has robotics mechanism to scan and trace automatically. For each subject, we performed the procedure 5 times for each applied pressure level. We could thus obtain 5 different radial pulse waveforms for the same person's same position at different applied pressures. All these processes were repeated twice for test reproducibility. Result: Aortic AI, peripheral AI and radial AI were higher in women than in men (P<0.01), radial AI strongly correlated with aortic AI, and radial AI was consistently approximately 39% higher than aortic AI. Relationship between representative radial AI of DMP-3000 and peripheral AI of SphygmoCor had strongly correlation. And there were three patterns in change of pulse waveform. Conclusion: In this study, it is revealed the new device was sufficient to measure how radial AI and radial waveform from the same person at the same time change under applied pressure and it had inverse-proportion to applied pressure.