Dependency parsing in Myanmar face many challenges due to agglutinative morphology, flexible word order, and lack of explicit word boundaries. This paper addresses these challenges by investigating the performance of Myanmar dependency parsing based three pretrained-transformer models: XLM-RoBERTa-Large, XLM-RoBERTa-base-Longformer-4096, and language-specific model, MyanBERTa. The proposed methodology utilizes a parameter-efficient model with adapter layers and a biaffine parsing mechanism by optimizing multi-task learning objective across six linguistic prediction tasks. The experimental results demonstrated that XLM-RoBERTa-Large achieves the highest Labeled Attachment Score (LAS) and Unlabeled Attachment Score (UAS) on both development set and test set. Despite XLM-RoBERTa-base-Longformer-4096 model can extended its input capacity and MyanBERTa is pretrained on a language capacity, their performance drop compared to XLM-RoBERTa-Large. These findings suggest that the scale and architectural strength of models are more critical for high-performance dependency parsing than models with an extended input capacity or those trained on a language-specific corpus.
Autonomous vehicles (AVs) increasingly incorporate artificial intelligence (AI) to improve perception, decision-making, and control. Integrating AI within AVs will have benefits, including reduced road deaths, lower emissions, and increased mobility, but failures have led to severe, sometimes fatal, outcomes. This paper develops a typology of AI failure modes in AVs, categorizing root causes into five categories: (1) biases in training data, (2) faulty algorithms, (3) out-of-distribution scenarios, (4) perceptual faults, and (5) adversarial attacks. We discuss the safety implications of each failure mode using literature, real-world case studies, and other examples. We propose several recommended mitigation strategies, including improved data practices, stronger model design, better sensing, and adversarial robustness. We also suggest future research avenues and make suggestions for our colleagues who are researchers, engineers, and policymakers to enhance the reliability and trustworthiness of AI for autonomous driving.
Light cryptography is a variant of conventional cryptography designed to minimize resource requirements. It is particularly well suited to the connected devices of the Internet of Things (IoT). Remote patient monitoring using the Internet of Things is a fast-moving field. This article presents a secure system for remotely monitoring vital patient parameters such as body temperature, heart rate and rhythm, while geolocating the patient. The device is placed on the patient to record his or her parameters, as well as geographical data, and then sends it to the doctor. Remote management of health data, such as vital parameters and patient geolocation, requires enhanced security due to the sensitivity of the information transmitted to healthcare professionals. Wireless communication networks are particularly vulnerable to security threats. This is why cryptography plays a crucial role in securing the data transmitted, from the sensor to the main node, before it is displayed online. In this context, the Ngrock platform is used to create a secure tunnel, remote local access to the module. A study of the most suitable and fastest algorithms was carried out, and we opted for the AES algorithm, recognized as one of the best algorithms thanks to its simplicity, lightness and suitability for IoT applications. At the end of the project, we produced a 3D-modelled prototype in line with the initial specifications, accompanied by an AES algorithm from a dedicated library, to ensure data security. Performance analysis showed that the algorithm used had no noticeable impact on system execution time, with an observed lag of around 1ms.
Land Surface Temperature (LST) is a key parameter for climate and environmental studies, typically derived from satellite thermal infrared observations. Accurate estimation of LST requires correction for atmospheric absorption and surface emissivity. This study compares the Split-Window Algorithm (SWA) and the Enterprise Algorithm (EA) for LST retrieval from the Visible Infrared Imaging Radiometer Suite (VIIRS) onboard the National Oceanic and Atmospheric Administration (NOAA)-20 satellite. Algorithm coefficients were derived from 135,000 synthetic cases generated with the MODTRAN 4.0 radiative transfer model, using Thermodynamic Initial Guess Retrieval (TIGR-2) atmospheric profiles and emissivity spectra from the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) library. Sensitivity analyses assessed the impact of sensor noise, emissivity uncertainty, and atmospheric water vapor. Results indicate that emissivity uncertainty dominates retrieval errors, with SWA achieving 1.25–1.64 K and EA 1.80–2.29 K. Validation at the Walpeup and Hay sites in Australia confirmed both methods perform consistently, with SWA slightly more robust under variable conditions.
The introduction of Artificial Intelligence (AI) into the domains of traditional art (visual arts, performing arts, and crafts) has sparked a complicated discussion about whether this might be an agent of disruption or an enhancement of our traditional art forms. This paper looks at the duality of AI, exploring the ways that recent technologies like Generative Adversarial Networks and Diffusion Models, and text-to-image generators are changing the fields of painting, sculpture, calligraphy, dance, music, and the arts of craft. Using examples and data, we illustrate the ways that AI can democratize creative expression, improve productivity, and preserve cultural heritage, while also examining the negative aspects, including: the threats to authenticity within art, ethical concerns around data, and issues including socio-economic factors such as job losses. While we argue for the context-dependence of the impact of AI (the potential for creative homogenization and the devaluation of human agency in artmaking), we also illustrate the potential for hybrid practices featuring AI in cuisine, etc. We advocate for the development of ethical guidelines, collaborative approaches, and inclusive technology development. In sum, we are articulating a vision of AI in which it amplifies our innate creativity while resisting the displacement of the cultural, nuanced, and emotional aspects of traditional art. The future will be determined by human choices about how to govern AI so that it becomes a mechanism for artistic evolution and not a substitute for the artist’s soul.
