Semiconductor ALD (Atomic Layer Deposition) is a precision-critical process involving sequential stages and large-scale time-series data from recipe settings and sensors. This paper proposes an edge AI architecture combining lightweight models on edge devices with centralized model training. The system enables early predictions from recipe data, real-time adjustments via sensor inputs, and continuous refinement using post-process outcomes. Only extracted features and result data (film thickness and uniformity) are transmitted to reduce communication overhead and protect sensitive data. The architecture supports performance monitoring and seamless model redeployment, adapting to changing equipment and environments. This approach improves product quality, reduces defect rates, and enhances manufacturing adaptability.

This paper presents the design and implementation of a system for processing and analyzing large-scale time-series data generated in semiconductor deposition processes. By adopting a real-time data collection and analysis architecture divided into Edge and Server layers, the system enables continuous retraining and updating of machine learning models based on real-time data streams. The evaluation of the model's performance demonstrates that additional training data significantly improves the model's accuracy in predicting process outcomes. Our approach not only provides a practical solution for real-time decision-making support in semiconductor manufacturing but also offers a scalable and adaptable framework applicable to various industrial sectors requiring real-time data analysis and processing. The results highlight the potential of integrating big data and artificial intelligence technologies to drive industrial innovation and optimize manufacturing processes.

While generative adversarial network (GAN) models have shown success in generating synthetic data of acoustic, image, and speech, research on generating seismic waves using GAN is receiving great attention. Although some methods have been successful in generating seismic data, they lack the ability to control the generated seismic waves according to earthquake parameters. This letter proposes a novel approach for controllable seismic wave synthesis using auxiliary classifier GAN (ACGAN). Our method focuses on the generation of synthetic seismic waveforms associated with earthquakes of different epicenteral distances. To incorporate distance information into our model, we introduce a distance regression loss function. In addition, we incorporate a feature-level diversity improvement regularization into our model to enhance the diversity of the generated seismic data. The proposed model was trained on KiK-net datasets, and the quality of the generated data was rigorously validated using various validation methods. Experimental results demonstrate the effectiveness of our proposed model in generating seismic waves by adjusting the earthquake epicenter distance.

Estimating earthquake parameters is an essential process for an earthquake analysis system. In particular, the magnitude and epicentral distance of an earthquake are the most basic parameters in earthquake analysis. To estimate these, the existing approaches require long waveform data from multiple stations. In this letter, we propose a novel estimation method based on multitasking deep learning and a convolutional recurrent neural network (CRNN) using only a single station. We also use the stream maximum of the input waveform to accurately estimate the earthquake magnitude. Based on the evaluation using the Stanford Earthquake dataset (STEAD) and the Kiban Kyoshin Network (KiK-net) dataset, we verify the high performance of the proposed method.

Image inpainting is an old problem in computer vision that restores occluded regions and completes damaged images. In the case of facial image inpainting, most of the methods generate only one result for each masked image, even though there are other reasonable possibilities. To prevent any potential biases and unnatural constraints stemming from generating only one image, we propose a novel framework for diverse facial inpainting exploiting the embedding space of StyleGAN. Our framework employs pSp encoder and SeFa algorithm to identify semantic components of the StyleGAN embeddings and feed them into our proposed SPARN decoder that adopts region normalization for plausible inpainting. We demonstrate that our proposed method outperforms several state-of-the-art methods.

In a next generation network (NGN) service environment, telecommunication service will continue to change from technology-oriented service to service-oriented service and customers will demand value-added Differentiated Services. To succeed in this business environment, service providers must support the evolution of next generation services and change their existing operational support system (OSS) into a highly distributed, loosely coupled and open-standard based OSS architecture. Also, OSS must shift to include business-oriented aspects. For the NGN telecommunication management, the Telemanagement (TM) Forum has developed a Next Generation Operational Systems and Software (NGOSS) architecture. One of the cornerstones of the NGOSS is the e-TOM business process model. In the service delivery management point of view, in the past, service delivery steps mostly depended on manual jobs and were based on single domain and single technology based-service delivery architecture. This has been a bottleneck, and has prevented flow-through and integrated service delivery in NGN service environment from being realized. An NGOSS-based automatic service delivery system using enterprise application integration (EAI) technology, including business process automation service using workflow, messaging service using a common messaging bus and XML messages, and connectivity service using transport protocol adaptor framework, will allow OSS developers and service providers to quickly automate business processes across the enterprise which are extendable to allow for new services quickly and easily.