Intel (United States)

Hallyu, or the Korean Wave, has evolved into a global cultural phenomenon that shapes consumption patterns, identity formation, and lifestyle orientations across national boundaries. In Indonesia, the extensive exposure of Generation Z to Korean popular culture through digital platforms positions this cohort as a strategically significant audience within transnational cultural flows. This study examines the threat and vulnerability of Hallyu exposure among Indonesian Generation Z and examines its implications for cultural resilience in a rapidly globalizing cultural environment. Employing a qualitative descriptive analytical approach, the study draws on non participatory observation, semi structured interviews with institutional stakeholders, cultural actors, and youth communities, and analysis of official cultural documents. Data were interpreted using Prunckun’s threat and vulnerability assessment framework and an intelligence informed analytical perspective. The findings indicate a moderate threat level with a total threat coefficient of 13, driven primarily by high capability factors, particularly knowledge and resource mobilization supporting Hallyu dissemination. The vulnerability assessment yields a moderate vulnerability profile (coefficient score: 9), reflecting strong attractiveness, facilitated digital penetration, and cumulative cultural impact. The study argues that Hallyu constitutes an invisible cultural security challenge that does not manifest through coercion, but through gradual normalization and asymmetry in cultural influence. Strengthening cultural resilience through governance, literacy, and competitive local cultural ecosystems is therefore essential to ensure balanced cultural development without securitizing popular cultural consumption.

System-on-chip security architecture is a critical, complex, and time-consuming activity, consuming months of effort. Furthermore, the architectural design can include subtle errors that compromise the security of the entire system. In this paper, we develop a security engine infrastructure, SEnTry , for systematically creating security architectures for protecting SoC designs against a variety of security subversions. SEnTry provides a plug-and-play, configurable subsystem composed of custom IPs that can be integrated into the platform to derive different security primitives. We develop an instance of SEnTry for supply-chain attacks. We discuss the spectrum of challenges involved in developing a unified architecture for systematic protection against the variety of attacks involved and the SEnTry approach to addressing them. We provide several case studies to demonstrate SEnTry design and perform extensive experiments to evaluate its overhead on multiple ASIC technologies. Our experiments suggest that SEnTry incurs minimal overhead in area and power consumption.

MotivationIn the global effort to characterize biodiversity, short species-specific genomic sequences known as DNA barcodes enable fine-grained comparisons among organisms within the same kingdom of life. Although machine learning algorithms specifically designed for the analysis of DNA barcodes are becoming more popular, most existing methodologies rely on generic supervised training algorithms.ResultsWe introduce BarcodeBERT, a family of models tailored to biodiversity analysis and trained exclusively on data from a reference library of 1.5 M invertebrate DNA barcodes. We evaluate BarcodeBERT on taxonomic identification tasks against a spectrum of machine learning approaches, including supervised training of classical neural architectures and fine-tuning of general DNA foundation models. Our self-supervised pretraining strategies on domain-specific data outperform fine-tuned foundation models, especially in identification tasks involving lower taxa such as genera and species. Compared with BLAST, a widely used sequence-search tool, BarcodeBERT achieves comparable species-level classification accuracy while being 55× faster. Our analysis of masking and tokenization strategies also provides practical guidance for building customized DNA language models, emphasizing the importance of aligning model training strategies with dataset characteristics and domain knowledge.Availability and implementationThe code repository is available at https://github.com/bioscan-ml/BarcodeBERT.

Local field potentials (LFPs) reflect coordination among neural populations, yet their exact relationship to neural computation remains unknown. One exception is the theta rhythm of the rodent hippocampus, which organizes sequential firing among place cells, enabling spike timing to track the animal's path through its environment. But when the animal stops, the theta rhythm becomes irregular, which is assumed to disrupt its ability to carry spatial information. Here we challenge this assumption by developing an artificial neural network that discovers position-tuned theta rhythms (pThetas) from LFPs even in the absence of strong theta oscillations. Using recordings from male rats, we provide evidence that pTheta is distinct from the dominant theta rhythm, while reflecting rhythmic coordination among place cell populations. Our work suggests that weak and intermittent oscillations, as seen in many brain regions and species, can convey information commensurate with population spike codes when decoded using information-based rather than variance-based principles.

Innovations in advanced GaN technology are shaping how power and data are delivered. The advent of 300 mm GaN‐on‐silicon is driving an inflection point in power and radio frequency (RF) electronics with more than 10X gains in performance figure‐of‐merits (FOMs) over silicon and III‐V technologies. The gains are made possible by advanced 300 mm process technologies that enable the scaling of GaN transistors to achieve higher transistor performance and density. As a result, the technology has led to fast‐switching and efficient power delivery solutions with ultrahigh densities, making it relevant for high‐performance compute and datacenter applications. Moreover, 3D monolithic integration of GaN and Si CMOS can be achieved on 300 mm GaN‐on‐silicon, enabling new possibilities, such as DrGaN for power electronics, and a fully integrated GaN mmwave RF beamformer for phased‐array applications. Progress has been made toward advancing 300 mm GaN‐on‐silicon for production with improvements in the tool ecosystem, the advent of engineered substrates, and better understanding of reliability.

We propose HiSum, a novel Hierarchical and temporally-enhanced framework for the task of long video Summarization. A significant challenge in this domain is generating concise yet informative summaries for long videos, which requires a nuanced understanding of both spatial and temporal information. Existing methods are often limited, struggling with temporally complex summaries and supporting videos typically shorter than one hour. To address these limitations, we propose HiSum, a training-free framework for long video summarization that is hierarchical, scene-aware, and temporally enhanced. HiSum will segment a video based on scene boundaries and then construct a summary hierarchically, effectively overcoming context length constraints. A dedicated temporal enhancement component is employed to mitigate information loss between chunks. Experimental results demonstrate that HiSum outperforms not only in resource consumption, but also in summary generation quality. Evaluated on the InfiniBench dataset, HiSum attained a GPT-5 score of 7.51 (on a 0-10 scale), significantly outperforming prevailing methods like Qwen2.5-VL, which scored 3.47.<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup>

The scaling of visual Artificial Intelligence (AI) for applications like smart cities and retail analytics is increasingly constrained by data pipeline inefficiencies rather than model computation. Conventional video codecs necessitate full-resolution decoding and extensive data movement, resulting in unnecessary power expenditure and GPU under-utilization. This paper evaluates MPEG-5 Low Complexity Enhancement Video Coding (LCEVC) [1] as a hierarchical solution that enables "base-first" operation, allowing AI inference to process only the low-resolution base stream. Using Intel’s Deep Learning Streamer framework on an integrated GPU architecture (Intel Core Ultra 7 258V), we quantify LCEVC’s impact in a high-density, eight Ultra High Definition (UHD) parallel AI inference stream environment compared to native AVC. Results confirm LCEVC-enhanced AVC achieves up to 3.4x pipeline speed-up, 71% lower average power consumption per channel, and a crucial 83% reduction in memory bandwidth utilization. These findings position LCEVC as a practical enabler for energy-efficient, high-density edge AI systems.

Understanding the dynamics of human gait, including both motions and forces, is vital to promote human mobility. While deep learning models may have advantages over costly laboratory-based experiments and physics-based simulations, existing models have been trained on small datasets with homogeneous demographics and focus on predicting a single output. We developed GaitDynamics, a generative foundation model trained on a large dataset of diverse gait patterns, which allows for flexible inputs, outputs and clinical applications. We illustrate the use of GaitDynamics for: (1) estimating ground reaction forces from kinematics with high accuracy even with missing kinematic data, (2) predicting the effects of gait modifications on knee loading without resource-intensive experiments and (3) predicting kinematic and force changes that occur with increasing running speeds. Our results demonstrate the accuracy and efficiency of GaitDynamics, showing its potential to assess and optimize gait for injury prevention, disease treatment and performance coaching. All data, code and trained models are publicly shared.