Найдено 480
Ничего не найдено
TagRecon: Fine-Grained 3D Reconstruction of Multiple Tagged Packages via RFID Systems
Wang Z., Duan C., Xue J., Li F., Feng Q., Zhu Y., Zhou Z.
Q1
To meet the new requirements of Industry 4.0, the logistics field has introduced 3D reconstruction technology. Computer vision-based solutions face challenges like bad lighting conditions and line-of-sight constraints. Meanwhile, the widespread adoption of RFID tags in supply chains offers an opportunity to enhance current reconstruction methods.
In this paper, we propose TagRecon, a fine-grained multi-object 3D reconstruction scheme utilizing well-deployed RFIDs. Specifically, TagRecon transforms the task of reconstruction into a problem of estimating 3D bounding boxes for tagged packages. By placing dual anchor tags on each target package, TagRecon enables accurate inference of the package’s translation and rotation using RFID-based localization and orientation sensing. Our scheme introduces a novel method to estimate rotations and translations for tagged packages, utilizing the known geometric relationship of anchor tags. Besides, to achieve simultaneous reconstruction of multiple packages, we manage to match tags from various packages through the correlation between anchor tag pairs. As far as we know, this is the first RFID-based solution that can simultaneously realize 3D translation and rotation estimation of multiple objects to a fine granularity. Experiments validate TagRecon achieves a 28.0 cm translation error and 6.8°, 6.0°, and 7.5° rotation errors for roll, pitch, and yaw angles on average.
Sarcasm Identification and Classification in Hindi Newspaper Headlines
Ahmad I., Gatla P., Mundotiya R.
Q2
ACM Transactions on Asian and Low-Resource Language Information Processing,
2025,
цитирований: 0,
doi.org,
Abstract
Sarcasm identification in textual data is the most captivating area of research in the current research trends. It is a challenging task for humans as well as for the computer. In this article, we have tried to identify sarcasm in the Hindi newspaper headlines of two of the most-read Hindi newspapers in India, namely Hindustan and Dainik Jagran. Initially, we collected 88,518 Hindi newspaper headlines and identified 1,945 headlines to be sarcastic, which we have considered for the present study. The headlines taken into consideration belong to the political domain and were published during some of the recent Legislative Assembly Elections of 2020, 2021, and 2022. Various machine learning and deep learning techniques have been used to develop the baseline models. It justifies the assumption that sarcastic text does not always bear a negative sentiment. It may bear a positive sentiment depending on the context. The present article aims at the creation of a dataset consisting of 1,945 Hindi newspaper headlines, training and testing machine learning and deep learning models, namely Extra Trees Classifier, Random Forest Classifier, XGBClassifier, fasttext-stackedTCN, and mBERT-stackedTCN for sarcasm identification on the dataset and comparing the results obtained by the models after the experiment. Out of all the choosen models, the Random Forest Classifier performs better with
\(F_1\)
score of 92.11 before data augmentation and 90.68 after data augmentation.
Evaluating Self-Supervised Learning for WiFi CSI-Based Human Activity Recognition
Xu K., Wang J., Zhu H., Zheng D.
Q1
With the advancement of the Internet of Things (IoT), WiFi Channel State Information (CSI)-based Human Activity Recognition (HAR) has garnered increasing attention from both academic and industrial communities. However, the scarcity of labeled data remains a prominent challenge in CSI-based HAR, primarily due to privacy concerns and the incomprehensibility of CSI data. Concurrently, Self-Supervised Learning (SSL) has emerged as a promising approach for addressing the dilemma of insufficient labeled data. In this paper, we undertake a comprehensive inventory and analysis of different categories of SSL algorithms, encompassing both previously studied and unexplored approaches within the field. We provide an in-depth investigation and evaluation of SSL algorithms in the context of WiFi CSI-based HAR, utilizing publicly available datasets that encompass various tasks and environmental settings. To ensure relevance to real-world applications, we design experiment settings aligned with specific requirements. Furthermore, our experimental findings uncover several limitations and blind spots in existing work, shedding light on the barriers that need to be addressed before SSL can be effectively deployed in real-world WiFi-based HAR applications. Our results also serve as practical guidelines and provide valuable insights for future research endeavors in this field.
A Survey of Document Stemming Algorithms in Information Retrieval Systems
Alyousf M., Alhalabi M.F.
Q2
ACM Transactions on Asian and Low-Resource Language Information Processing,
2025,
цитирований: 0,
doi.org,
Abstract
With the increase in the growth and diversity of databases and the enormity of their contents, there has become an urgent need to find advanced techniques in Natural Language Processing (NLP) applications, especially in the field of Information Retrieval (IR). One of the most popular techniques that can improve information retrieval is the stemming of text documents. Given the importance of stemming for information retrieval systems, in this paper, we present a detailed study of the adopted stemming approaches and the working mechanism of the various algorithms that follow each approach. We analyzed and evaluated the most important algorithms by comparing them based on specific criteria, including their strength in stemming, their advantages, and the disadvantages of each. Based on this comparison, we can identify the weaknesses that each stemming algorithm suffers from. We mainly aim through the study that we conducted in this paper to try to overcome the weaknesses of these algorithms and take advantage of their most important advantages to develop a new more efficient stemming algorithm for the English language.
RoboCam: Model-Based Robotic Visual Sensing for Precise Inspection of Mesh Screens
Zhou S., Le D.V., Jiang L., Chen Z., Peng X., Ho D., Zheng J., Tan R.
Q1
The 3D-printed mesh screen with dense penetrating pores is a new structure for massive manufacturing of molded pulp package products. However, some of the pores may be clogged by the printing material powder during the printing process. Such defects negatively affect the quality of the pulp packages produced using the mesh screen mold. To pinpoint the defects, we design a model-based robotic visual sensing system, called RoboCam, which uses a robotic arm to carry a high-resolution camera for full inspection of a mold consisting of joined mesh screens. To inspect the entire mold, RoboCam plans the camera poses to capture multiple images of the mold and render synthesized images as references for identifying the clogged pores. In particular, we propose novel designs to rectify the inherent run-time pose errors of the robotic system for ensuring the reference quality and to accelerate the reference rendering for reducing inspection latency. Extensive evaluation shows that RoboCam’s design outperforms various baselines, including three existing computer vision and convolution neural network-based inspection systems. RoboCam achieves a recall rate of 94.95% within 528 seconds latency for inspecting an entire mold with 13,000 designed pores.
Artificial Intelligence as a Service (AIaaS) for Cloud, Fog and the Edge: State-of-the-Art Practices
Syed N., Anwar A., Baig Z., Zeadally S.
Q1
Artificial Intelligence (AI) fosters enormous business opportunities that build and utilize private AI models. Implementing AI models at scale and ensuring cost-effective production of AI-based technologies through entirely in-house capabilities is a challenge. The success of the Infrastructure as a Service (IaaS) and Software as a Service (SaaS) Cloud Computing models can be leveraged to facilitate a cost-effective and scalable AI service paradigm, namely, ‘AI as a Service.’ We summarize current state-of-the-art solutions for AI-as-a-Service (AIaaS), and we discuss its prospects for growth and opportunities to advance the concept. To this end, we perform a thorough review of recent research on AI and various deployment strategies for emerging domains considering both technical as well as survey articles. Next, we identify various characteristics and capabilities that need to be met before an AIaaS model can be successfully designed and deployed. Based on this we present a general framework of an AIaaS architecture that integrates the required aaS characteristics with the capabilities of AI. We also compare various approaches for offering AIaaS to end users. Finally, we illustrate several real-world use cases for AIaaS models, followed by a discussion of some of the challenges that must be addressed to enable AIaaS adoption.
A Hybrid Approach for Localisation of Sensor Nodes in Remote Locations
Hada R.P., Srivastava A.
Q1
A Wireless Sensor Network (WSN) is a network of sensor nodes using low-power wireless technology to collect data in a region of interest (ROI). Due to their low energy, locating sensor nodes in large outdoor areas is challenging, which precludes GPS integration. WSNs typically comprise a small number of beacon nodes (BN) whose locations are known in advance, with most nodes deployed at unknown coordinates within the ROI. Endeavours to determine the locations of such unknown WSN nodes are largely based on the impractical assumption that every unknown node (UN) is within the communication range of BNs. Subsequently, these approaches utilise at least two BNs to determine the position of one UN. The Received Signal Strength Indicator (RSSI) or Angle of Arrival (AoA) values of the signals from the BNs form the basis for such localisation. This article suggests an iterative hybrid approach incorporating AoA and RSSI techniques, achieving accurate localisation with just one BN. The iterative method gradually covers the region, avoiding the unrealistic assumption of having all UNs within range. It also presents an innovative use of a unipolar stepper motor for AoA measurements. Experiments in a simulated environment and a real-world prototype validate the approach’s effectiveness.
Bidirectional Directed Acyclic Graph Neural Network for Aspect-level Sentiment Classification
Xu J., Xiao L., Wu A., Ma T., Dong D., He L.
Q2
ACM Transactions on Asian and Low-Resource Language Information Processing,
2025,
цитирований: 0,
doi.org,
Abstract
To achieve outstanding aspect-level sentiment analysis (ASC), it is crucial to reduce the distance between aspect terms and opinion words. Recently, advanced methods in ASC used graph neural network (GNN)-based methods to leverage the syntactic dependency within the sentence, which can shorten the distance through syntactical dependencies. However, existing approaches that utilize GNNs have difficulty extracting long-distance relations in the dependency tree due to the over-smoothing problem resulting from stacking GNN layers, which limits their ability to detect remote relations. To solve this issue, we propose a Bidirectional Directed Acyclic Graph (BDAG) to reconstruct syntactic dependencies and a Bidirectional Directed Acyclic Graph Neural Network (BDAGNN) to efficiently propagate multi-hop sentiment information. We also enhance the BDAG with affective commonsense knowledge from SenticNet for comprehensive sentiment classification. The BDAGNN we proposed obtains partial state-of-the-art performance on four benchmark datasets, indicating the feasibility of encoding syntactic structures with BDAG.
Contextualized Quaternion Embedding Towards Polysemy in Knowledge Graph for Link Prediction
Chen J., Wang Y., Zhao S., Zhou P., Zhang Y.
Q2
ACM Transactions on Asian and Low-Resource Language Information Processing,
2025,
цитирований: 0,
doi.org,
Abstract
To meet the challenge of incompleteness within Knowledge Graphs, Knowledge Graph Embedding(KGE) has emerged as the fundamental methodology for predicting the missing link(Link Prediction), by mapping entities and relations as low-dimensional vectors in continuous space. However, current KGE models often struggle with the polysemy issue, where entities exhibit different semantic characteristics depending on the relations in which they participate. Such limitation stems from weak interactions between entities and their relation contexts, leading to low expressiveness in modeling complex structures and resulting in inaccurate predictions. To address this, we propose
ConQuatE
(
Con
textualized
Quat
ernion
E
mbedding), a model that enhances the representation learning of entities across multiple semantic dimensions by leveraging quaternion rotation to capture diverse relational contexts. In specific, ConQuatE incorporates contextual cues from various connected relations to enrich the original entity representations. Notably, this is achieved through efficient vector transformations in quaternion space, without any extra information required other than original triples. Experimental results demonstrate that our model outperforms state-of-the-art models for Link Prediction on four widely-recognized datasets: FB15k-237, WN18RR, FB15k and WN18.
RPWAEAuth: Sensor-Based Continuous Authentication Using Reconstruction Probability in Wasserstein Autoencoder
Li Y., Wang Y., Huang H.
Q1
Nowadays, with the widespread adoption of mobile devices, information security has become particularly important. Existing sensor-based continuous authentication systems ensure the security of mobile devices to some extent, but most have drawbacks, such as lacking end-to-end structure or requiring data from both legitimate users and imposters for training. In this paper, we present RPWAEAuth, a sensor-based continuous Authentication system using Reconstruction Probability in the Wasserstein AutoEncoder. RPWAEAuth implicitly collects user behavior patterns from the built-in accelerometer, gyroscope, and magnetometer of mobile devices. The Wasserstein autoencoder maps the sensor data into a continuous latent space close to a prior distribution and reconstructs them using reconstruction probability for better authentication. In the registration stage, RPWAEAuth collects and preprocesses the sensor data from a legitimate user for RPWAE training. In the authentication stage, when a user interacts with the device, RPWAEAuth collects and preprocesses the sensor data, and then feeds them into the trained RPWAE to generate a reconstruction probability. This probability is then compared with a predefined threshold for user authentication. We evaluate the performance of RPWAEAuth on our dataset in terms of the effectiveness of RPWAEAuth, impact of sensor numbers, effectiveness of reconstruction probability, authentication time, resilience to mimic attacks, comparison with different AEs, and comparison with state-of-the-art methods. The experimental results demonstrate that RPWAEAuth achieves superior authentication performance compared to other methods, with an accuracy of 99.34% and an EER of 0.66% on 69 unseen users.
Enabling Low-Power Massive MIMO with Ternary ADCs for AIoT Sensing
Liu S., Fu N.
Q1
The proliferation of networked devices and the surging demand for ubiquitous intelligence have given rise to the artificial intelligence of things (AIoT). However, the utilization of high-resolution analog-to-digital converters (ADCs) and numerous radio frequency chains significantly raises power consumption. This paper explores a cost-effective solution using ternary ADCs (T-ADCs) in massive multiple-input-multiple-output (MIMO) systems for low-power AIoT and specifically addresses channel sensing challenges. The channel is first estimated through a pilot-aided scheme and refined using a joint-pilot-and-data (JPD) approach. To assess the performance limits of this two-threshold ADC system, the analysis includes its hardware-ideal counterpart, the parallel one-bit ADCs (PO-ADCs) and a realistic scenario where noise variance is unknown at the receiver is considered. Analytical findings indicate that the JPD scheme effectively mitigates performance degradation in channel estimation due to coarse quantization effects under mild conditions, without necessitating additional pilot overhead. For deterministic and random channels, we propose modified expectation maximization (EM) and variational inference EM estimators, respectively. Extensive simulations validate the theoretical results and demonstrate the effectiveness of the proposed estimators in terms of mean square error and symbol error rate, which showcases the feasibility of implementing T-ADCs and the associated JPD scheme for greener AIoT smart sensing.
Exploring Semantic Attributes for Image Caption Synthesis in Low-Resource Assamese Language
Choudhury P., Guha P., Nandi S.
Q2
ACM Transactions on Asian and Low-Resource Language Information Processing,
2025,
цитирований: 0,
doi.org,
Abstract
Research on image caption generation has predominantly focused on resource-rich languages like English, leaving resource-poor languages (like Assamese and several others) largely understudied. In this context, this paper leverages both visual and semantic attribute based features for generating captions in Assamese language. Semantic attributes refer to the significant words that represent higher-level knowledge about the image content. This work contributes through the effective use of features derived from semantic words in low resource Assamese language. The second contribution is the proposal of a Visual-Semantic Self-Attention (VSSA) module for the combination of features derived from images and semantic attributes. The VSSA module enables the image captioning model to dynamically attend to relevant regions of the image as well as the important semantic attributes, thereby leading to more contextually relevant and linguistically accurate Assamese captions. Moreover, the VSSA module is incorporated into a Transformer model to leverage the stacked attention for performance improvement. The model is trained by using both cross-entropy loss optimization and reinforcement learning approach. The effectiveness of the proposed model is evaluated through both qualitative and quantitative analyses (using BLEU-n and CIDEr metrics). The proposed model shows significant performance improvement in Assamese caption synthesis compared to previous methods, achieving 93.7% CIDEr score on the COCO-Assamese Caption (COCO-AC) dataset.
Facial Expression Analysis in Parkinson's Disease Using Machine Learning: A Review
Camargo G., Ngo Q., Passos L., Jodas D., Papa J., Kumar D.
Q1
Computerised facial expression analysis is performed for a range of social and commercial applications and more recently its potential in medicine such as to detect Parkinson’s Disease (PD) is emerging. This has possibilities for use in telehealth and population screening. The advancement of facial expression analysis using machine learning is relatively recent, with a majority of the published work being post-2019. We have performed a systematic review of the English-based publication on the topic from 2019 to 2024 to capture the trends and identify research opportunities that will facilitate the translation of this technology for recognising Parkinson’s disease. The review shows significant advancements in the field, with facial expressions emerging as a potential biomarker for PD. Different machine learning models, from shallow to deep learning, could detect PD faces. However, the main limitation is the reliance on limited datasets. Furthermore, while significant progress has been made, model generalization must be tested before clinical applications.
Transfer Learning in Sensor-Based Human Activity Recognition: A Survey
Dhekane S.G., Ploetz T.
Q1
Sensor-based human activity recognition (HAR) has been an active research area for many years, resulting in practical applications in smart environments, assisted living, fitness, healthcare, and more. Recently, deep-learning-based end-to-end training has pushed the state-of-the-art performance in domains such as computer vision and natural language, where large amounts of annotated data are available. However, large quantities of annotated data are typically not available for sensor-based HAR. Moreover, the real-world settings on which HAR is performed differ in terms of sensor modalities, classification tasks, and target users. To address this problem, transfer learning has been explored extensively. In this survey, we focus on these transfer learning methods in the application domains of smart home and wearables-based HAR. In particular, we provide a
problem–solution
perspective by categorizing and presenting the works in terms of their contributions and the challenges they address. We present an overview of the state of the art for both application domains. Based on our analysis of 246 papers, we highlight the gaps in the literature and provide a roadmap for addressing these. This survey provides a reference to the HAR community by summarizing the existing works and providing a promising research agenda.
MixSong: Diverse and Strictly Formatted Chinese Poetry Generation
Song X., Song C., Yu H., Zhu Y., Yao H.
Q2
ACM Transactions on Asian and Low-Resource Language Information Processing,
2025,
цитирований: 0,
doi.org,
Abstract
Chinese poetry, renowned for its elegance and simplicity, is a hallmark of Chinese culture. While neural networks have made significant advancements in generating poetry, balancing diversity with adherence to rigid structural formats remains a challenge. Research indicates that factors such as themes, emotions (e.g., happiness, sadness), and sentiments (e.g., positive, negative) play a crucial role in poetic creation, influencing both the diversity and quality of the generated content. In this paper, we propose MixSong, an autoregressive language model based on the Transformer architecture, designed to incorporate a wide range of conditional factors. MixSong utilizes adversarial training to integrate these factors, enabling the model to implicitly learn distributional information in the latent space. Additionally, we introduce several uniquely customized symbol sets, including paragraph identifiers, position identifiers, rhyme identifiers, tune identifiers, and conditional distinctive identifiers. These symbols help MixSong effectively capture and enforce the constraints necessary for generating high-quality poetry. Extensive experimental results demonstrate that MixSong significantly outperforms existing models in both automatic metrics and human evaluations, achieving notable improvements in both diversity and quality of the generated poetry.
TransQAM: Transformer-based Question Answering System in Malayalam
Rahmath K R., Raj P.C R., P.C R.
Q2
ACM Transactions on Asian and Low-Resource Language Information Processing,
2025,
цитирований: 0,
doi.org,
Abstract
Question Answering (QA) systems are used to extract the exact answer from a given context. In this study, we have implemented a QA system named TransQAM with BERT and its variants for the low-resource Malayalam language. We have considered the transformer models, namely, BERT, multilingual BERT, XLM-RoBERTa, and MuRIL for implementation. Since there is no publicly available Malayalam dataset for QA, we have built and made publicly available a sufficiently large Malayalam QA dataset in SQuAD (Stanford Question Answering Dataset) format with 30k question-answer pairs. We have obtained state-of-the-art results for TransQAM implemented using MuRIL. Due to the advancement of language models and active research, many languages, such as English, have well-developed QA systems compared to the unexplored Malayalam language. According to our knowledge, TransQAM is the first QA system in Malayalam that successfully applies transformer models to answer questions and achieves more than 80% accuracy.
New Bagging Based Ensemble Learning Algorithm Distinguishing Short and Long Texts for Document Classification
Wang Y., Feng L.
Q2
ACM Transactions on Asian and Low-Resource Language Information Processing,
2025,
цитирований: 0,
doi.org,
Abstract
To improve the classification accuracy of ensemble learning, a new bootstrap aggregating (Bagging) ensemble learning algorithm distinguishing short and long texts for document classification is proposed. First, the performances of different typical deep learning methods on processing long and short texts are compared, and the optimal base classifiers for long and short texts are selected respectively. Second, the random sampling method in traditional bagging classification algorithms is improved, and a threshold group based random sampling method which can balance the numbers of long and short text subsets is proposed. Moreover, to improve the model inference speed and classification accuracy, the training of long and short text subsets is realized by combining the knowledge distillation theory. Finally, the sample classification probabilities on different categories are considered, and the category similarity information is combined with the traditional weighted voting classifier ensemble method to avoid the problem that the sampling process may decrease the accuracy. Experimental results on multiple datasets show that the algorithm can effectively improve the accuracy of document classification and has obvious advantages over typical deep learning algorithms and ensemble learning algorithms.
FLog: Automated Modeling of Link Quality for LoRa Networks in Orchards
Yang K., Chen Y., Du W.
Q1
LoRa networks have been deployed in many orchards for environmental monitoring and crop management. An accurate propagation model is essential for efficiently deploying a LoRa network in orchards, e.g., determining gateway coverage and sensor placement. Although some propagation models have been studied for LoRa networks, they are not suitable for orchard environments, because they do not consider the shadowing effect on wireless propagation caused by the ground and tree canopies. This article presents
FLog
, a propagation model for LoRa signals in orchard environments.
FLog
leverages a unique feature of orchards, i.e., all trees have similar shapes and are planted regularly in space. We develop a three-dimensional model of orchards. Once we have the location of a sensor and a gateway, we know the media that the wireless signal traverses. Based on this knowledge, we generate the First Fresnel Zone (FFZ) between the sender and the receiver. The intrinsic path loss exponents of all media can be combined into a classic Log-Normal Shadowing model in the FFZ. Extensive experiments in almond orchards show that
FLog
reduces the link quality estimation error by 42.7% and improves gateway coverage estimation accuracy by 70.3%, compared with a widely used propagation model. The source codes and dataset are released at
https://github.com/ycucm/Flog
.
Out-of-Distribution Data: An Acquaintance of Adversarial Examples - A Survey
Karunanayake N., Gunawardena R., Seneviratne S., Chawla S.
Q1
Deep neural networks (DNNs) deployed in real-world applications can encounter out-of-distribution (OOD) data and adversarial examples. These represent distinct forms of distributional shifts that can significantly impact DNNs’ reliability and robustness. Traditionally, research has addressed OOD detection and adversarial robustness as separate challenges. This survey focuses on the intersection of these two areas, examining how the research community has investigated them together. Consequently, we identify two key research directions: robust OOD detection and unified robustness. Robust OOD detection aims to differentiate between in-distribution (ID) data and OOD data, even when they are adversarially manipulated to deceive the OOD detector. Unified robustness seeks a single approach to make DNNs robust against both adversarial attacks and OOD inputs. Accordingly, first, we establish a taxonomy based on the concept of distributional shifts. This framework clarifies how robust OOD detection and unified robustness relate to other research areas addressing distributional shifts, such as OOD detection, open set recognition, and anomaly detection. Subsequently, we review existing work on robust OOD detection and unified robustness. Finally, we highlight the limitations of the existing work and propose promising research directions that explore adversarial and OOD inputs within a unified framework.
A Comprehensive Review on IoT Marketplace Matchmaking: Approaches, Opportunities and Challenges
An Q., Jiang F., Neiat A., Yeoh W., Venayagamoorthy K., Zaslavsky A.
Q1
Service discovery matchmaking plays a vital role in the cyber marketplace for the Internet of Things (IoT), especially in peer-to-peer environments where buyers and sellers dynamically register and match resource profiles online. As the IoT marketplace expands, efficient resource allocation through matchmaking is increasingly important. However, the growing complexity of service discovery, coupled with data security and privacy challenges, complicates the identification of suitable services. To address these issues, this study conducts a comprehensive review of matchmaking algorithms within the IoT marketplace by examining their key attributes, strengths, and limitations as documented in academic literature. This paper categorises and summarises state-of-the-art approaches, identifying research gaps and proposing future directions. Our comparative analysis highlights the strengths and weaknesses of current methodologies, advocating for deep learning and context-aware solutions to improve service efficiency. Additionally, blockchain-based approaches are discussed for their potential to improve security, trust, and privacy-preserving transactions. This research lays a critical foundation for the advancement of secure, efficient IoT-enabled marketplaces.
A Review on Edge Large Language Models: Design, Execution, and Applications
Zheng Y., Chen Y., Qian B., Shi X., Shu Y., Chen J.
Q1
Large language models (LLMs) have revolutionized natural language processing with their exceptional understanding, synthesizing, and reasoning capabilities. However, deploying LLMs on resource-constrained edge devices presents significant challenges due to computational limitations, memory constraints, and edge hardware heterogeneity. This survey provides a comprehensive overview of recent advancements in edge LLMs, covering the entire lifecycle — from resource-efficient model design and pre-deployment strategies to runtime inference optimizations. It also explores on-device applications across various domains. By synthesizing state-of-the-art techniques and identifying future research directions, this survey bridges the gap between the immense potential of LLMs and the constraints of edge computing.
Machine Learning for Infectious Disease Risk Prediction: A Survey
Liu M., Liu Y., Liu J.
Q1
Infectious diseases place a heavy burden on public health worldwide. In this article, we systematically investigate how machine learning (ML) can play an essential role in quantitatively characterizing disease transmission patterns and accurately predicting infectious disease risks. First, we introduce the background and motivation for using ML for infectious disease risk prediction. Next, we describe the development and application of various ML models for infectious disease risk prediction, categorizing them according to the models’ alignment with vital public health concerns specific to two distinct phases of infectious disease propagation: (1) the pandemic and epidemic phases (the P-E phases) and (2) the endemic and elimination phases (the E-E phases), with each presenting its own set of critical questions. Subsequently, we discuss challenges encountered when dealing with model inputs, designing task-oriented objectives, and conducting performance evaluations. We conclude with a discussion of open questions and future directions.
A Hybrid Statistical and Rule-based Approach to Extremely Low-resource Machine Transliteration
Connor P.C.
Q2
ACM Transactions on Asian and Low-Resource Language Information Processing,
2025,
цитирований: 0,
doi.org,
Abstract
Machine transliteration work has focused primarily on languages with large volumes of parallel corpus, and between language pairs whose orthographies are very different. In contrast, a large proportion of the world’s languages have vastly fewer resources and employ Roman-like alphabets often with large degrees of orthographic overlap with high-resource languages. We propose that machine transliteration between languages with few training examples can be accomplished by a noisy-channel-like statistical model captured in a human editable format with practical rule-based capabilities built-in. This hybrid approach allows users to take advantage of an algorithm to find and apply common transformations in context while providing rigorous control over the output. Effectiveness is evaluated on the Bible names translation matrix dataset of Wu et al. (2018), covering 591 languages that involve 590 names on average per language pair. Our approach slightly exceeds past results and explores several features targeted at benefiting the extremely low-resource language domain.
