The use of NLP in the realm of financial technology is broad and complex, with applications ranging from sentiment analysis and named entity recognition to question answering. Large Language Models (LLMs) have been shown to be effective on a variety of tasks; however, no LLM specialized for the financial domain has been reported in literature. In this work, we present BloombergGPT, a 50 billion parameter language model that is trained on a wide range of financial data. We construct a 363 billion token dataset based on Bloomberg's extensive data sources, perhaps the largest domain-specific dataset yet, augmented with 345 billion tokens from general purpose datasets. We validate BloombergGPT on standard LLM benchmarks, open financial benchmarks, and a suite of internal benchmarks that most accurately reflect our intended usage. Our mixed dataset training leads to a model that outperforms existing models on financial tasks by significant margins without sacrificing performance on general LLM benchmarks. Additionally, we explain our modeling choices, training process, and evaluation methodology. We release Training Chronicles (Appendix C) detailing our experience in training BloombergGPT.
This monograph provides an overview of general deep learning methodology and its applications to a variety of signal and information processing tasks. The application areas are chosen with the following three criteria in mind: (1) expertise or knowledge of the authors; (2) the application areas that have already been transformed by the successful use of deep learning technology, such as speech recognition and computer vision; and (3) the application areas that have the potential to be impacted significantly by deep learning and that have been experiencing research growth, including natural language and text processing, information retrieval, and multimodal information processing empowered by multi-task deep learning.
Information technology (IT) allows members of the growing elderly population to remain independent longer. However, while technology becomes more and more pervasive, an age-related underutilisation of IT remains observable. For instance, elderly people (65 years of age and older) are significantly less likely to use the Internet than the average population (see, for instance, European Commission, 2011). This age-related digital divide prevents many elderly people from using IT to enhance their quality of life through tools, such as Internet-based service delivery. Despite the significance of this phenomenon, the information systems (IS) literature lacks a comprehensive consideration and explanation of technology acceptance in general and more specifically, Internet adoption by the elderly. This paper thus studies the intentions of the elderly with regard to Internet use and identifies important influencing factors. Four alternative models based on technology acceptance theory are tested in the context of comprehensive survey data. As a result, a model that explains as much as 84% of the variance in technology adoption among the elderly is developed. We discuss the contribution of our analyses to the research on Internet adoption (and IT adoption in general) by the elderly, on the digital divide, and on technology acceptance and identify potentially effective paths for future research and theoretical development.
Meenakshi Sudarvizhi Seenipeyathevar, Prasath Palaniappan, Vijayakumar Arumugam
et al.
ABSTRACT This study deals with an integrated experimental‐machine learning framework for wear estimation in functionally graded composites made from recycled magnesium machining chips, using low‐cost ceramic fibers as reinforcement with the radial Modeling technique. The primary hurdle that is being addressed is the accurate prediction of wear behavior in spatially graded magnesium matrix composites, while simultaneously avoiding extensive experimental testing. Under varying degrees of applied loads (4.4 to 39 N), sliding speeds (0.45 to 4.5 m/s), and sliding distances (500 to 4500 m), the wear performance was experimentally assessed. Results demonstrate a hardness increment of 26.26% in the outer region compared to the inner region, while resistance to wear was enhanced by 19.8% in the outer zone due to the grading of ceramic fibers. A limited experimental dataset consisting of wear measurements from the inner, middle, and outer zones of the composite was utilized in developing and validating four machine‐learning models for wear rate prediction. The tree‐based ensemble methods significantly outperformed deep‐learning strategies, with the LightGBM model providing the best prediction performance across all zones and achieving optimization with a maximum tree depth of 5, 480 leaves, and a feature fraction of 0.05. Moreover, zone‐specific XGBoost models were also developed, employing customized learning rates and minimal loss reduction parameters in order to elevate prediction accuracy. The proposed machine‐learning framework thus provides a pathway for rapid and reliable wear rate estimation for ceramic fiber‐reinforced magnesium composites, significantly lessening experimental burden. Results highlight that recycled magnesium waste, when combined with ceramic reinforcement, can be effectively employed to produce sustainable and economically viable materials with improved wear resistance, particularly for automotive and industrial applications.
To improve the counting accuracy in dense rice seed scenarios, this study proposes a YOLOv5-based dense rice seed counting method that integrates C3CBAM and Soft-NMS. This method integrates the CBAM attention module into the shallow C3 modules of the backbone network to enhance image features. Additionally, it removes the original large and medium-sized object detection heads of YOLOv5 and adds a dedicated detection head for tiny rice seeds. For post-processing of model prediction data, the Soft-NMS algorithm is employed to replace standard Non-Maximum Suppression (NMS) and reduce missed detections. Finally, image acquisition, seed counting, and a user interface are integrated into a single system, enabling rice breeders to conduct seed counting tasks more intuitively and efficiently. Compared with the baseline YOLOv5 model, the recall and mAP@[0.5:0.95] of the improved model increase by 6.4 % and 5.7 %, respectively. Furthermore, this study designs experiments with three levels of seed density. In the intermediate-type rice seed samples, the detection accuracy reaches 100 % under light and moderate density conditions, while it maintains stable counting performance under heavy density conditions with an accuracy above 99.7 %. This work significantly enhances rice seed counting efficiency for researchers and facilitates rice variety improvement studies.
Accurate estimation of tuna catch is crucial for effective pelagic fishery management and resource conservation. However, existing manual counting methods suffer from issues such as low accuracy and poor timeliness, highlighting the urgent need for an efficient and automated solution. This paper proposes an automatic tuna counting method based on the YOLOv8n-DMTNet target detection algorithm combined with the improved ByteTrack tracking algorithm. The method uses YOLOv8n as the base model, enhanced with detail-enhanced convolution and a multi-scale feature fusion pyramid network, which significantly improves detection accuracy in complex marine environments. Additionally, a dynamic, task-aligned detection head is introduced to optimize the synergy between classification and localization tasks. To further improve counting accuracy, the ByteTrack algorithm is employed for target tracking, and a region-specific counting method is designed to prevent double counting and omission due to occlusion and motion irregularities. Experimental results show that the improved YOLOv8n-DMTNet model achieves a 9.2% increase in mAP@0.5 and a 6.4% increase in mAP@0.5:0.95 compared to YOLOv8n in the tuna detection task, while reducing the number of parameters by 42.3% and computational complexity by 33.3%. The counting accuracy reaches 93.5%, and the method demonstrates superior performance in terms of accuracy, robustness, and computational resource efficiency, making it well-suited for resource-constrained fishing vessel environments. This approach provides reliable technical support for automated catch counting in pelagic fisheries.
The extraction of shape features from vector elements is essential in cartography and geographic information science, supporting a range of intelligent processing tasks. Traditional methods rely on different machine learning algorithms tailored to specific types of line and polygon elements, limiting their general applicability. This study introduces a novel approach called “Pre-Trained Shape Feature Representations from Transformers (PSRT)”, which utilizes transformer encoders designed with three self-supervised pre-training tasks: coordinate masking prediction, coordinate offset correction, and coordinate sequence rearrangement. This approach enables the extraction of general shape features applicable to both line and polygon elements, generating high-dimensional embedded feature vectors. These vectors facilitate downstream tasks like shape classification, pattern recognition, and cartographic generalization. Our experimental results show that PSRT can extract vector shape features effectively without needing labeled samples and is adaptable to various types of vector features. Compared to the methods without pre-training, PSRT enhances training efficiency by over five times and improves accuracy by 5–10% in tasks such as line element matching and polygon shape classification. This innovative approach offers a more unified, efficient solution for processing vector shape data across different applications.
This study presents a Historic Building Information Modeling (HBIM)-based approach for the digital restoration and documentation of lost wooden architectural heritage. The approach was applied to Building 1-2 of Hyeumwonji, the site of a temporary Goryeo Dynasty palace in Paju, South Korea. To reconstruct this lost structure, we combined historical and archaeological analyses to estimate the original design and generated blueprints that guided the HBIM-based 3D model of the building. We collected LiDAR point cloud data from the site, aligned them with the HBIM model, and visualized the integrated result using Unreal Engine 5. The outcome was a comprehensive virtual restoration comprising 13,814 individual building elements. This case study demonstrates that, even with minimal physical remains, wooden heritage sites can be digitally restored by leveraging HBIM and historical reasoning. It also highlights the strengths of HBIM in version tracking, incorporation of historical updates, and systematic documentation throughout the restoration process. Compared to traditional 2D CAD-based restoration methods, the HBIM approach offers significant advantages in terms of updatability, data integration, and long-term preservation of restoration data. Overall, the project illustrates how combining rigorous historical analysis with advanced digital modeling can revive lost heritage architecture in virtual form, providing a rich resource for research and conservation.
In recent years, silicon photomultipliers (SiPMs) have emerged as preferred photoelectric conversion devices in positron emission tomography (PET) due to their outstanding performance. SiPMs possess single-photon resolution capability and time resolution below 100 ps, enabling precise photon arrival time measurements. These advances paved the way for emerging applications such as time-of-flight PET (TOF-PET), photon counting CT, and positron emission lifetime imaging, presenting new challenges to SiPM performance, the advancing of which to their physical limits has become a key focus area in next-generation SiPM research. In traditional SiPM architectures, signal processing and analog-to-digital conversion introduce noise and degrade time performance, thereby limiting the full SiPM potential. With the recent and rapid development of semiconductor manufacturing processes, SiPMs could be manufactured on standard CMOS process nodes, which marks a significant breakthrough in the SiPM field, allowing for the integration of digital logic within SiPM devices. This advancement opens the possibility of achieving more precise time, energy, and position information within a single SiPM, thereby providing potential possibilities to push SiPMs to their performance limits. In this study, we reviewed the development history, working principles, and performance parameters of SiPMs. We analyzed the limitations of traditional SiPMs, outlined key aspects of digital SiPM research, and introduced various current digital SiPM architectures. Finally, we summarized and anticipated key technologies in digital SiPMs.
One relevant aspect in the development of the Semantic Web framework is the achievement of a real inter-agents communication capability at the semantic level. The agents should be able to communicate and understand each other using standard communication protocols freely, that is, without needing a laborious a priori preparation, before the communication takes place. For that setting we present in this paper a proposal that promotes to describe standard communication protocols using Semantic Web technology (specifically, OWL-DL and SWRL). Those protocols are constituted by communication acts. In our proposal those communication acts are described as terms that belong to a communication acts ontology, that we have developed, called CommOnt. The intended semantics associated to the communication acts in the ontology is expressed through social commitments that are formalized as fluents in the Event Calculus. In summary, OWL-DL reasoners and rule engines help in our proposal for reasoning about protocols. We define some comparison relationships (dealing with notions of equivalence and specialization) between protocols used by agents from different systems.
A modification to the vis-viva equation that accounts for general relativistic effects is introduced to enhance the accuracy of predictions of orbital motion and precession. The updated equation reduces to the traditional vis-viva equation under Newtonian conditions and is a more accurate tool for astrodynamics than the traditional equation. Preliminary simulation results demonstrate the application potential of the modified vis-viva equation for more complex n-body systems. Spherical symmetry is assumed in this approach; however, this limitation could be removed in future research. This study is a pivotal step toward bridging classical and relativistic mechanics and thus makes an important contribution to the field of celestial dynamics.