Artificial-intelligence (AI) technology is used in a variety of fields, from robot cleaner motion control to call center counselors, AI speakers, and Mars exploration. Because the technology levels of all applications and services that utilize AI vary widely, it is not possible to view all applications using AI technology at the same level. Nevertheless, there have been no cases in which the level of AI technology was defined. Therefore, the Electronics and Telecommunications Research Institute (ETRI) Artificial Intelligence Research Laboratory has defined the levels of the main technical elements of AI from steps 1 to 6. In this report, the Artificial Intelligence Technology Level 1.0 (ATL 1.0) is presented. It was established by comprehensively referring to the AI technology prospects and technology roadmaps of major countries. It is hoped that it can be used as a measure for determining the levels of AI applications or services or as an indicator for establishing a technology roadmap.

Recently, a technique for applying a deep learning language model pretrained from a large corpus to fine-tuning for each application task has been widely used as a language processing technology. The pretrained language model shows higher performance and satisfactory generalization performance than existing methods. This paper introduces the major research trends related to deep learning pretrained language models in the field of language processing. We describe in detail the motivations, models, learning methods, and results of the BERT language model that had significant influence on subsequent studies. Subsequently, we introduce the results of language model studies after BERT, focusing on SpanBERT, RoBERTa, ALBERT, BART, and ELECTRA. Finally, we introduce the KorBERT pretrained language model, which shows satisfactory performance in Korean language. In addition, we introduce techniques on how to apply the pretrained language model to Korean (agglutinative) language, which consists of a combination of content and functional morphemes, unlike English (refractive) language whose endings change depending on the application.

Temporal action detection (TAD) in untrimmed videos is an important but a challenging problem in the field of computer vision and has gathered increasing interest recently. Although most studies on action in videos have addressed action recognition in trimmed videos, TAD methods are required to understand real-world untrimmed videos, including mostly background and some meaningful action instances belonging to multiple action classes. TAD is mainly composed of temporal action localization that generates temporal action proposals, such as single action and action recognition, which classifies action proposals into action classes. However, the task of generating temporal action proposals with accurate temporal boundaries is challenging in TAD. In this paper, we discuss TAD technologies that are considered high performance in terms of representative TAD studies based on deep learning. Further, we investigate evaluation methodologies for TAD, such as benchmark datasets and performance measures, and subsequently compare the performance of the discussed TAD models.

This paper examines the trends of recently developed human-care robots and social interaction technologies. As one of the solutions to the problems of an aging society, human-care robots have gained considerable attention from the public and the market. However, commercialized human-care robots do not meet user expectations for the role as companions. Current robot services based on short-term interaction and fragmentary pieces of intelligence have encountered difficulty in eliciting natural communication with humans. This results in the failure of human-robot social bonding. Social interaction is being actively investigated as a technique for improving robots' natural communication skills. Robots can form a natural bond with humans through social interaction, which consequently increases their effectiveness. In this paper, we introduce recent human-care robot-related issues and subsequently describe technical challenges and implications for the success of human-care robots. In addition, we review recent trends on social interaction technologies and the datasets required.

The cloud computing industry has focused on establishing a cloud-based business environment for enterprises with efforts to convert using their own on-premise computing infrastructures to using cloud services. With these efforts, using cloud services has become natural, especially for the IT industry. The cloud computing industry is moving toward proliferation of the cloud computing environment into various evolving industries. Along with industrial trends, new technical trends such as edge computing and multi-cloud are emerging. These trends are expected to create new business models and develop related service ecosystems, providing new opportunities for service providers and new experiences for users. A mong those emerging technologies, multi-cloud technology is expected to realize unlimited global cloud computing resources by unifying cloud resources from multiple public cloud service providers. In this paper, we introduce the concept and related trends of multi-cloud technology. Subsequently, we analyze the main functionalities and several use cases of multi-cloud technology. Finally, we summarize the effects and usefulness of multi-cloud technology in the domestic cloud industry.

With the development of nonvolatile memory technology, Intel has released the Optane datacenter persistent memory module (DCPMM) that can be deployed in the dual in-line memory module. The results of research and experiments on Optane DCPMMs are significantly different from the anticipated results in previous studies through emulation. The DCPMM can be used in two different modes, namely, memory mode (similar to volatile DRAM: Dynamic Random Access Memory) and app direct mode (similar to file storage). It has buffers in 256-byte granularity; this is four times the CPU (Central Processing Unit) cache line (i.e., 64 bytes). However, these properties are not easy to use correctly, and the incorrect use of these properties may result in performance degradation. Optane has the same characteristics of DRAM and storage devices. To take advantage of the performance characteristics of this device, operating systems and applications require new approaches. However, this change in computing environments will require a significant number of researches in the future.

As the increasing expectations of a practical AI (Artificial Intelligence) service makes AI algorithms more complicated, an efficient processor to process AI algorithms is required. To meet this requirement, processors optimized for parallel processing, such as GPUs (Graphics Processing Units), have been widely employed. However, the GPU has a generalized structure for various applications, so it is not optimized for the AI algorithm. Therefore, research on the development of AI processors optimized for AI algorithm processing has been actively conducted. This paper briefly introduces an AI processor especially for inference acceleration, developed by the Electronics and Telecommunications Research Institute, South Korea., and other global vendors for mobile and server platforms. However, the GPU has a generalized structure for various applications, so it is not optimized for the AI algorithm. Therefore, research on the development of AI processors optimized for AI algorithm processing has been actively conducted.

Neuromorphic hardware refers to brain-inspired computers or components that model an artificial neural network comprising densely connected parallel neurons and synapses. The major element in the widespread deployment of neural networks in embedded devices are efficient architecture for neuromorphic hardware with regard to performance, power consumption, and chip area. Spiking neural networks (SiNNs) are brain-inspired in which the communication among neurons is modeled in the form of spikes. Owing to brainlike operating modes, SNNs can be power efficient. However, issues still exist with research and actual application of SNNs. In this issue, we focus on the technology development cases and market trends of two typical tracks, which are listed above, from the point of view of artificial intelligence neuromorphic circuits and subsequently describe their future development prospects.

In the maritime industry, the convergence of information and communication technology (ICT) occurs in two areas. The first one is the convergence of e-navigation technology to apply ICT to the maritime domain. This was led by EU's initiative and is also being promoted in our country. The other one is the convergence of autonomous ship technology to apply various ICT technologies for commercial purposes. This initiative is being led by maritime industries. In this paper, we describe trends of e-navigation and maritime autonomous surface ship technology convergence. Subsequently, we explain the standardization status of performance and technical standard bodies, international standardization organizations, and related standard-developing organizations to provide the interoperability of these technologies. Finally, we examine ongoing projects in Korea and suggest future directions for their R&D and standardization.

Precise detection and modulation of electrical signal patterns passing through peripheral nerves connecting organs and brainstems, referred to as electroceuticals or bioelectronic medicines, have emerged as a new type of treatments for neural disorders and chronic diseases. With the rapid advancements in neural interface technologies, electroceuticals are the focus of treatments for these disorders or diseases. In this paper, we introduced electroceuticals as an extension of neuromodulation for the treatment of chronic diseases, such as diabetes, rheumatoid arthritis, obesity, and bladder dysfunction, without side effects that are unavoidably elicited when conventional drugs are taken. Further, this paper reviewed the anatomy of the peripheral nervous system, treatment examples for chronic diseases, technological demands for peripheral nerve interfacing, global R&D programs and market trends for electroceuticals, and prospects on electroceuticals.