Search

The fast evolution of artificial intelligence frameworks has resulted in the creation of increasingly sophisticated large language models (LLM), ChatGPT being the most famous one. This study paper dives into this LLM with a case study of ChatGPT's architecture and provides a thorough comparative analysis of its numerous versions, tracking its history from its conception to its most recent incarnations. This research intends to give a full knowledge of the model's history by investigating the underlying mechanisms and enhancements provided in each edition. The comparative analysis covers key aspects such as model size, training data, fine-tuning techniques, and performance metrics. Furthermore, this study evaluates the limits of ChatGPT in its many incarnations. These limitations include common sense reasoning difficulties, biased replies, verbosity, sensitivity to input wording, and others. Each constraint is investigated for potential remedies and workarounds. This research article also provides a complete analysis of the ChatGPT architecture and its progress through multiple iterations. It gives vital insights for academics, developers, and users wanting to harness the promise of ChatGPT while managing its restrictions by exploring both the model's strengths and limitations. The distinctiveness of this paper rests in its comprehensive assessment of ChatGPT's architectural development and its practical strategy for resolving the myriad difficulties in producing cohesive and contextually relevant replies. [ABSTRACT FROM AUTHOR]

Over the last decade, a plethora of training datasets have been compiled for use in language-based machine perception and in human-centered AI, alongside research regarding their compilation methods. From a primarily linguistic perspective, we add to these studies in two ways. First, we provide an overview of sixty-six training datasets used in automatic image, video, and audio captioning, examining their compilation methods with a metadata analysis. Second, we delve into the annotation process of crowdsourced datasets with an interest in understanding the linguistic factors that affect the form and content of the captions, such as contextualization and perspectivation. With a qualitative content analysis, we examine annotator instructions with a selection of eleven datasets. Drawing from various theoretical frameworks that help assess the effectiveness of the instructions, we discuss the visual and textual presentation of the instructions, as well as the perspective-guidance that is an essential part of the language instructions. While our analysis indicates that some standards in the formulation of instructions seem to have formed in the field, we also identified various reoccurring issues potentially hindering readability and comprehensibility of the instructions, and therefore, caption quality. To enhance readability, we emphasize the importance of text structure, organization of the information, consistent use of typographical cues, and clarity of language use. Last, engaging with previous research, we assess the compilation of both web-sourced and crowdsourced captioning datasets from various perspectives, discussing factors affecting the diversity of the datasets. [ABSTRACT FROM AUTHOR]

Early skin cancer diagnosis saves lives as the disease can be successfully treated through complete excision. Computer-aided diagnosis methods are developed using artificial intelligence techniques to help earlier detection and identify hidden causes leading to cancers in skin lesion images. In skin cancer image classification problems, an ensemble of classifiers has demonstrated better classification ability than a single classification algorithm. Traditionally, training an ensemble uses the complete set of original features, where some of these features can be redundant or irrelevant and hence, may not provide useful information in generating good models for ensemble classification. Moreover, newly created features may help improve classification performance. To address this issue, the existing methods have used feature construction for building an ensemble classifier, which usually creates a fixed number of features that may fit the training data too well, resulting in poor test performance. This study develops a novel classification approach that combines ensemble learning, feature selection, and feature construction utilizing genetic programming (GP) to handle the above limitations. The proposed method automatically evolves variable-length feature vectors consisting of GP-selected and GP-constructed features suitable for training an ensemble classifier. This study evaluates the effectiveness of the proposed method on two benchmark real-world skin image datasets that include dermoscopy and standard camera images. The experimental results reveal that the proposed algorithm significantly outperforms four state-of-the-art convolutional neural network methods, the existing GP approaches, and 11 commonly used machine learning methods. Furthermore, this study also includes interpreting evolved individuals that highlight important skin cancer characteristics playing a vital role in discriminating images of different cancer classes. This study shows that high classification performance can be achieved at a low cost of computational resources and inference time, and accordingly, this method is potentially suitable to be implemented in mobile devices for the automated screening of skin lesions and many other malignancies in low-resource settings. [ABSTRACT FROM AUTHOR]

Large variations across images make image classification a challenging task; limited training data further increases its difficulty. Genetic programming (GP) has been considerably applied to image classification. However, most GP methods tend to directly evolve a single classifier or depend on a predefined classification algorithm, which typically does not lead to ideal generalization performance when only a few training instances are available. Applying ensemble learning to classification often outperforms employing a single classifier. However, single-tree representation (each individual contains a single tree) is widely employed in GP. Training multiple diverse and accurate base learners/classifiers based on single-tree GP is challenging. Therefore, this article proposes a new ensemble construction method based on multi-tree GP (each individual contains multiple trees) for image classification. A single individual forms an ensemble, and its multiple trees constitute base learners. To find the best individual in which multiple trees are diverse and effectively cooperate, i.e., the nth tree can correct the errors of the previous n-1 trees, the new method assigns different weights to multiple trees using the idea of AdaBoost and performs classification via weighted majority voting. Furthermore, a new tree representation is developed to evolve diverse and accurate base learners that extract useful features and conduct classification simultaneously. The new approach achieves significantly better performance than almost all benchmark methods on eight datasets. Additional analyses highlight the effectiveness of the new ensembles and tree representation, demonstrating the potential for providing valuable interpretability in ensemble trees. [ABSTRACT FROM AUTHOR]

Despite the recent advancements in dialogue systems, persona-driven chatbots are still in their infancy. Previous studies on persona-driven dialogue generation demonstrated its ability in generating responses that contain more detailed persona information. However, the challenge of maintaining persona consistency and contextual coherence still persists in persona-driven dialogue generation. Moreover, current methods have limitations in processing multi-source inputs and identifying interlocutor intents due to the absence of trustworthy labels and effective modeling. Additionally, numerous approaches rely on pre-trained large-scale language models that require costly computational resources. To address these challenges, a lightweight hierarchical intent-inferring pointer network is proposed for multi-source persona-driven dialogue generation. The proposed method involves detecting interlocutor intents in chitchat and utilizing pseudo labeling and natural language inference techniques to generate intent labels. Our model is evaluated on a benchmark dataset PersonaChat. The experimental results show that our model outperforms the strongest baseline by 13.47% and 4.28% in terms of persona consistency and contextual coherence, respectively. [ABSTRACT FROM AUTHOR]

The intelligent soft robot has received wide attention from both academia and the industry due to its remarkable adaptability. It performs intelligent behavioral learning and evolved morphologies in unpredictable environmental conditions. However, designing a soft robot with a well-adapted morphology involves searching through a large number of possible structures. Furthermore, to learn control tasks in diverse environments, the robot performs computationally intensive numerical simulations, which is time-consuming for evaluating the performance of robots. To address both issues, a multi-fidelity evolutionary algorithm is proposed, which consists of three main components. Firstly, a niching-based fidelity adjustment strategy is introduced to significantly reduce the evaluation cost by training the controller of each robot for only a small number of simulation steps. In particular, considering the estimation errors of the low-fidelity evaluation, the population is divided into multiple subpopulations with different fidelity levels for parallel optimization. Secondly, an effective morphology transfer strategy is proposed to improve the quality of offspring by transferring the local structure of robots in different subpopulations. Finally, a fast local search is developed to enhance the search efficiency of the algorithm without performing additional control simulations. The experimental results on 31 test tasks demonstrate that the proposed algorithm outperforms the SOTA design algorithms on 25 test tasks, especially when the computational budget is limited. Compared to the baseline algorithms, our algorithm reduces the computational cost by 60$\%$% while achieving similar performance. [ABSTRACT FROM AUTHOR]

Abstract Background Diagnoses entered by general practitioners into electronic medical records have great potential for research and practice, but unfortunately, diagnoses are often in uncoded format, making them of little use. Natural language processing (NLP) could assist in coding free-text diagnoses, but NLP models require local training data to unlock their potential. The aim of this study was to develop a framework of research-relevant diagnostic codes, to test the framework using free-text diagnoses from a Swiss primary care database and to generate training data for NLP modelling. Methods The framework of diagnostic codes was developed based on input from local stakeholders and consideration of epidemiological data. After pre-testing, the framework contained 105 diagnostic codes, which were then applied by two raters who independently coded randomly drawn lines of free text (LoFT) from diagnosis lists extracted from the electronic medical records of 3000 patients of 27 general practitioners. Coding frequency and mean occurrence rates (n and %) and inter-rater reliability (IRR) of coding were calculated using Cohen’s kappa (Κ). Results The sample consisted of 26,980 LoFT and in 56.3% no code could be assigned because it was not a specific diagnosis. The most common diagnostic codes were, ‘dorsopathies’ (3.9%, a code covering all types of back problems, including non-specific lower back pain, scoliosis, and others) and ‘other diseases of the circulatory system’ (3.1%). Raters were in almost perfect agreement (Κ ≥ 0.81) for 69 of the 105 diagnostic codes, and 28 codes showed a substantial agreement (K between 0.61 and 0.80). Both high coding frequency and almost perfect agreement were found in 37 codes, including codes that are particularly difficult to identify from components of the electronic medical record, such as musculoskeletal conditions, cancer or tobacco use. Conclusion The coding framework was characterised by a subset of very frequent and highly reliable diagnostic codes, which will be the most valuable targets for training NLP models for automated disease classification based on free-text diagnoses from Swiss general practice.

Background: Diagnoses entered by general practitioners into electronic medical records have great potential for research and practice, but unfortunately, diagnoses are often in uncoded format, making them of little use. Natural language processing (NLP) could assist in coding free-text diagnoses, but NLP models require local training data to unlock their potential. The aim of this study was to develop a framework of research-relevant diagnostic codes, to test the framework using free-text diagnoses from a Swiss primary care database and to generate training data for NLP modelling. Methods: The framework of diagnostic codes was developed based on input from local stakeholders and consideration of epidemiological data. After pre-testing, the framework contained 105 diagnostic codes, which were then applied by two raters who independently coded randomly drawn lines of free text (LoFT) from diagnosis lists extracted from the electronic medical records of 3000 patients of 27 general practitioners. Coding frequency and mean occurrence rates (n and %) and inter-rater reliability (IRR) of coding were calculated using Cohen's kappa (Κ). Results: The sample consisted of 26,980 LoFT and in 56.3% no code could be assigned because it was not a specific diagnosis. The most common diagnostic codes were, 'dorsopathies' (3.9%, a code covering all types of back problems, including non-specific lower back pain, scoliosis, and others) and 'other diseases of the circulatory system' (3.1%). Raters were in almost perfect agreement (Κ ≥ 0.81) for 69 of the 105 diagnostic codes, and 28 codes showed a substantial agreement (K between 0.61 and 0.80). Both high coding frequency and almost perfect agreement were found in 37 codes, including codes that are particularly difficult to identify from components of the electronic medical record, such as musculoskeletal conditions, cancer or tobacco use. Conclusion: The coding framework was characterised by a subset of very frequent and highly reliable diagnostic codes, which will be the most valuable targets for training NLP models for automated disease classification based on free-text diagnoses from Swiss general practice. [ABSTRACT FROM AUTHOR]

Facial expression recognition (FER) attracts much attention in computer vision. Previous works mostly study the single-label FER problem. The more complex multi-label facial expression recognition task is underexplored. Multi-label FER is more challenging than single-label task due to two primary causes. On one hand, there are less available multi-label facial expression data for analysis. On the other hand, the entanglement of expressions makes recognition more difficult. In this work, we leverage class activation map (CAM) to improve the performance of multi-label FER. Firstly, considering the shortage of training data, an attention flipping consistency (AFC) loss equipped with random rotation augmentation is proposed. It restrains the network to produce consistent CAMs under horizontally flipping transformation and thus improves the stability of network without any extra data. Secondly, based on the prior knowledge that different emotions have different predominant activated facial regions, we propose a label-guided spatial attention dispersing (SAD) loss to enable the model to learn from distinct expression-related regions. By combining the widely used multi-label classification loss (i.e., binary cross-entropy loss) and proposed AFC loss and SAD loss, our method achieves state-of-the-art performance on multi-label FER databases and the model’s interpretability is improved. [ABSTRACT FROM AUTHOR]

Generative adversarial networks (GANs) have emerged as a powerful solution for generating synthetic data when the availability of large, labeled training datasets is limited or costly in large-scale machine learning systems. Recent advancements in GAN models have extended their applications across diverse domains, including medicine, robotics, and content synthesis. These advanced GAN models have gained recognition for their excellent accuracy by scaling the model. However, existing accelerators face scalability challenges when dealing with large-scale GAN models. As the size of GAN models increases, the demand for computation and communication resources during inference continues to grow. To address this scalability issue, this article proposes Chiplet-GAN, a chiplet-based accelerator design for GAN inference. Chiplet-GAN enables scalability by adding more chiplets to the system, thereby supporting the scaling of computation capabilities. To handle the increasing communication demand as the system and model scale, a novel interconnection network with adaptive topology and passive/active network links is developed to provide adequate communication support for Chiplet-GAN. Coupled with workload partition and allocation algorithms, Chiplet-GAN reduces execution time and energy consumption for GAN inference workloads as both model and chiplet-system scales. Evaluation results using various GAN models show the effectiveness of Chiplet-GAN. On average, compared to GANAX, SpAtten, and Simba, the Chiplet-GAN reduces execution time and energy consumption by 34% and 21%, respectively. Furthermore, as the system scales for large-scale GAN model inference, Chiplet-GAN achieves reductions in execution time of up to 63% compared to the Simba, a chiplet-based accelerator. [ABSTRACT FROM AUTHOR]

This research introduces an algorithmic framework for enhancing the security of Internet of Things (IoT) networks. The Enhanced Anomaly Detection (EAD) algorithm initiates the process by detecting anomalies in real-time IoT data, serving as the foundational layer. The Behavior Analysis for Profiling (BAP) algorithm builds upon EAD, adding behavior analysis for profiling and adaptive identification of abnormal behavior. Signature-Based Detection (SBD) involves pre-identified attack signatures, which supports detection of known attacks and provides proactive defense measures against documented threats. The MLID, or the Machine Learning-Based Intrusion Detection, algorithm uses trained machine learning models in order to detect anomalies and the adaptability to changing security risks. The Real-Time Threat Intelligence Integration (RTI) algorithm integrates updated threat intelligence feeds, which improves the framework's responsiveness to emerging threats. The visual representations illustrate once again the idea of the new framework being very accurate at intergration, applicability, and overal security effectiveness. The research makes a standard solution which proves to be a smart and responsive way guarding the IoT networks reducing and even fighting known and potential threats in a real-time mode. [ABSTRACT FROM AUTHOR]

Federated learning (FL) has emerged as one of the de-facto privacy-preserving paradigms that can effectively work with decentralized data sources (e.g., hospitals) without acquiring any private data. Recently, applications of FL have vastly expanded into multiple domains, particularly the medical domain, and FL is becoming one of the mainstream technologies of the near future. In this study, we provide insights into FL fundamental concepts (e.g., the difference from centralized learning, functions of clients and servers, workflows, and nature of data), architecture and applications in the general medical domain, synergies with emerging technologies, key challenges (medical domain), and potential research prospects. We discuss major taxonomies of the FL systems and enlist technical factors in the FL ecosystem that are the foundation of many adversarial attacks on these systems. We also highlight the promising applications of FL in the medical domain by taking the recent COVID-19 pandemic as an application use case. We highlight potential research and development trajectories to further enhance the persuasiveness of this emerging paradigm from the technical point of view. We aim to concisely present the progress of FL up to the present in the medical domain including COVID-19 and to suggest future research trajectories in this area. [ABSTRACT FROM AUTHOR]

ABSTRACTThe current land cover (LC) mapping paradigm relies on automatic satellite imagery classification, predominantly through supervised methods, which depend on training data to calibrate classification algorithms. Hence, training data have a critical influence on classification accuracy. Although research on specific aspects of training data in the LC classification context exists, a study that organizes and synthetizes the multiplicity of aspects and findings of these researches is needed. In this article, we review the training data used for LC classification of satellite imagery. A protocol of identification and selection of relevant documents was followed, resulting in 114 peer-reviewed studies included. Main research topics were identified and documents were characterized according to their contribution to each topic, which allowed uncovering subtopics and categories and synthetizing the main findings regarding different aspects of the training dataset. The analysis found four research topics, namely construction of the training dataset, sample quality, sampling design and advanced learning techniques. Subtopics included sample collection method, sample cleaning procedures, sample size, sampling method, class balance and distribution, among others. A summary of the main findings and approaches provided an overview of the research in this area, which may serve as a starting point for new LC mapping initiatives.

Generative Artificial Intelligence (AI) systems, like ChatGPT, have the potential to perpetuate and amplify cultural biases embedded in their training data, which are predominantly produced by dominant cultural groups. This paper explores the philosophical and technical challenges of detecting and mitigating cultural bias in generative AI, drawing on Plato's Allegory of the Cave to frame the issue as a problem of limited and distorted representation. We propose a multifaceted approach combining technical interventions, such as data diversification and culturally aware model constraints, with a deeper engagement with the cultural and philosophical dimensions of the problem. Drawing on theories of extended cognition and situated knowledge, we argue that mitigating AI biases requires a reflexive interrogation of the cultural contexts of AI development and a commitment to empowering marginalized voices and perspectives. We claim that controlling cultural bias in generative AI is inseparable from the larger project of promoting equity, diversity, and inclusion in AI development and governance. By bridging philosophical reflection with technical innovation, this paper contributes to the growing discourse on responsible and inclusive AI, offering a roadmap for detecting and mitigating cultural biases while grappling with the profound cultural implications of these powerful technologies. [ABSTRACT FROM AUTHOR]

We study the stability of accuracy during the training of deep neural networks (DNNs). In this context, the training of a DNN is performed via the minimization of a cross-entropy loss function, and the performance metric is accuracy (the proportion of objects that are classified correctly). While training results in a decrease of loss, the accuracy does not necessarily increase during the process and may sometimes even decrease. The goal of achieving stability of accuracy is to ensure that if accuracy is high at some initial time, it remains high throughout training. A recent result by Berlyand, Jabin, and Safsten introduces a doubling condition on the training data, which ensures the stability of accuracy during training for DNNs using the absolute value activation function. For training data in R n , this doubling condition is formulated using slabs in R n and depends on the choice of the slabs. The goal of this paper is twofold. First, to make the doubling condition uniform, that is, independent of the choice of slabs. This leads to sufficient conditions for stability in terms of training data only. In other words, for a training set T that satisfies the uniform doubling condition, there exists a family of DNNs such that a DNN from this family with high accuracy on the training set at some training time t 0 will have high accuracy for all time t > t 0 . Moreover, establishing uniformity is necessary for the numerical implementation of the doubling condition. We demonstrate how to numerically implement a simplified version of this uniform doubling condition on a dataset and apply it to achieve stability of accuracy using a few model examples. The second goal is to extend the original stability results from the absolute value activation function to a broader class of piecewise linear activation functions with finitely many critical points, such as the popular Leaky ReLU. [ABSTRACT FROM AUTHOR]

Text-based emotion prediction is an important task in the field of affective computing. Most prior work has been restricted to predicting emotions corresponding to a few high-level emotion classes. This paper explores and experiments with various techniques for fine-grained (27 classes) emotion prediction $^\dagger$ † . In particular, 1) we present a method to incorporate multiple annotations from different raters, 2) we analyze the model's performance on fused emotion classes and with sub-sampled training data, 3) we present a method to leverage the correlations among the emotion categories, and 4) we propose a new framework for text-based fine-grained emotion prediction through emotion definition modeling. The emotion definition-based model outperforms the existing state-of-the-art for fine-grained emotion dataset GoEmotions. The approach involves a multi-task learning framework that models definitions of emotions as an auxiliary task while being trained on the primary task of emotion prediction. We model definitions using masked language modeling and class definition prediction tasks. We show that this trained model can be used for transfer learning on other benchmark datasets in emotion prediction with varying emotion label sets, domains, and sizes. The proposed models outperform the baselines on transfer learning experiments demonstrating the model's generalization capability. [ABSTRACT FROM AUTHOR]

Generative adversarial networks (GANs) are powerful deep learning models for synthesizing realistic data. However, their performance critically depends on curating optimal training data. This research conducts a comprehensive study analyzing the impact of sample size, class balance, and heterogeneity in training datasets on GAN image and text generation quality. Through extensive experiments on CIFAR-10, it has been demonstrated that insufficient samples, imbalanced classes, and lack of diversity cause degraded sample quality, coherence, and mode collapse. The analysis conducted in this research work provides unique insights into data-GAN interplay. Models trained on balanced subsets with adequate samples per class produce superior Inception Scores and BLEU, avoiding limited variety in outputs. The techniques presented enable developing more generalizable and creative GANs. This work proves to be the first of its kind to rigorously evaluate the role of data characteristics like size, balance and heterogeneity in stabilizing GAN training and improving output fdelity across modalities. The data-centric findings would be valuable for researchers to curate optimal datasets that can unlock GANs' full potential for diverse, realistic generation with wide applications in graphics, vision, language and beyond. [ABSTRACT FROM AUTHOR]

In countries like India, diseases in plants are a major concern in the agricultural sector. Crop loss due to diseases led to reduction in the quality and quantity of agricultural products. This also leads to economic losses. Hence, timely monitoring of plants is necessary. But monitoring diseases in large fields is a difficult task. To overcome this problem, effective management strategies should be taken to control diseases in plants. It can be done by acquiring data for disease identification and automation of the recognition of diseases. Deep learning is of great use in this area using the principles of object detection. In this paper, we are using YOLOv3 (you only look once) to determine the diseases in plants, based on the symptoms seen on the leaves. The advantage of using YOLOv3 is that multiple diseases can be detected on the image of a single leaf. An important feature of YOLOv3 is that it can detect small disease spots seen on the leaves. Here, we are concentrating on the bacterial spot disease seen on the bell pepper plant. The identification results show a mean average precision of 90%. [ABSTRACT FROM AUTHOR]

Artificial intelligence (AI) shows good potential for detecting and discriminating faults in electrical machines, however, they require initial training with sufficient data, which is almost impossible to collect for working electrical machines in the field. This paper proposes an effective approach to solve this problem by getting the required training data from exact simulation results. To evaluate this idea, the finite elements method is used to simulate a three-phase induction motor (IM) in the healthy state as well as the stator inter-turn fault, broken rotor bar fault, and mixed eccentricity fault conditions. Then, for every fault condition, some fault indices are extracted from the stator line current and used to arrange and train a suitable support vector machine (SVM) model to detect and discriminate the fault condition. A similar IM is prepared in the laboratory, where, its stator line currents are sampled and recorded under the healthy and the fault conditions, and the same fault indices are extracted from the stator currents. Some penalties, which are determined by comparing experimental test results and corresponding simulation results in the healthy state, are applied to the experimentally attained values of the indices. The modified indices are then applied to the trained SVM models, where, the attained results confirm the trained SVM models are equally able to detect and discriminate the faults in the real IMs. [ABSTRACT FROM AUTHOR]

Enzyme–substrate interactions play a fundamental role in elucidating synthesis pathways and synthetic biology, as they allow for the understanding of important aspects of a reaction. Establishing the interaction experimentally is a slow and costly process, which is why this problem has been addressed using computational methods such as molecular dynamics, molecular docking, and Monte Carlo simulations. Nevertheless, this type of method tends to be computationally slow when dealing with a large search space. Therefore, in recent years, methods based on artificial intelligence, such as support vector machines, neural networks, or decision trees, have been implemented, significantly reducing the computing time and covering vast search spaces. These methods significantly reduce the computation time and cover broad search spaces, rapidly reducing the number of interacting candidates, as they allow repetitive processes to be automated and patterns to be extracted, are adaptable, and have the capacity to handle large amounts of data. This article analyzes these artificial intelligence-based approaches, presenting their common structure, advantages, disadvantages, limitations, challenges, and future perspectives. [ABSTRACT FROM AUTHOR]

Identifying and extracting urban features such as buildings and producing accurate information about the location of features from remotely sensed images is important in updating maps, spatial databases, urban planning, and meeting the needs of urban services. Supervised methods of producing training data were widely applied in deriving accurate information from such images. However, those approaches require human intervention and are therefore time-consuming, and associated with classification accuracy limitations. This study presents a novel approach for extracting buildings from remotely sensed images by automating training data selection steps for supervised classification. To prepare training samples intelligently and automatically, a knowledge-based integration of radar and optical images was used. First, a general height constraint was applied to a digital surface model (DSM) obtained from radar images to divide the image pixels into two parts. Then, by applying a local height filter in each part, pixels representing buildings were extracted. Considering that the buildings are not the only features with significant heights, another constraint defined as coherence constraint was considered. Then, histogram of the optical image pixels corresponding to the filtered radar DSM pixels extracted from the previous steps was used to select sample training pixels. By considering pixels within the maximum range of the histogram, the final building samples for Support Vector Machine (SVM) classification were derived and applied for building extraction. The output showed that it sufficiently and reliably improved the existing supervised classification limitation to extract building. The performance of the proposed method was evaluated over different image datasets regardless of input image type and region. It achieved an overall average accuracy of 92% and a kappa coefficient of 0.83. [ABSTRACT FROM AUTHOR]

Purpose: The purpose of this commentary is to outline some of the pitfalls when using the fitness–fatigue model to unravel the interaction between training load and performance. By doing so, we encourage sport scientists and coaches to interpret the parameters from the model with some extra caution. Conclusions: Caution is needed when interpreting the fitness–fatigue model since the parameter values are influenced by the starting parameter values, the modeling technique, and the input of the model. Also, the use of general constants should be avoided since they do not account for interindividual differences and differences between training-load methods. Therefore, we advise sport scientists and coaches to use the model as a way to work more data-informed rather than working data-driven. [ABSTRACT FROM AUTHOR]

Federated learning (FL) is considered a de facto standard for privacy preservation in AI environments because it does not require data to be aggregated in some central place to train an AI model. Preserving data on the client side and sharing only the model’s parameters with a central server preserves privacy while training an AI model of higher generalizability. Unfortunately, sharing the model’s parameters with the server can create privacy leaks, and therefore, FL is unable to meet privacy requirements in many situations. Furthermore, FL is prone to other technical issues, such as data poisoning, model poisoning, fairness, client dropout, and convergence issues, to name just a few. In this work, we provide a multifaceted survey on FL, including its fundamentals, paradigm shifts, technical issues, recent developments, and future prospects. First, we discuss the fundamental concepts of FL (workflow, categorization, the differences between centralized learning and FL, and applications of FL in diverse fields), and we then discuss the paradigm shifts brought on by FL from a broader perspective (e.g., data use, AI model development, resource sharing, etc.). Later, we pinpoint ten practical issues currently hindering the viability of the FL landscape, and we discuss developments made under each issue by summarizing state-of-the-art (SOTA) literature. We highlight FL partnerships with two or more technologies that either improve practical aspects/issues in FL or extend its adoption to new areas/domains. We pinpoint various trade-offs that exist in an FL ecosystem, and the corresponding SOTA developments to mitigate them. We also discuss the latest studies that have been proposed to make FL trustworthy and beneficial for the community. Lastly, we suggest valuable research directions towards enhancing technical efficacy by guiding researchers to less explored topics in FL.

An increasing availability of remote sensing data in the era of geo big-data makes producing well-represented, reliable training data to be more challenging and requires an excessive amount of human labor. In addition, the rapid increase in graphics processing unit processing power has enabled the development of advanced deep learning algorithms, which achieve impressive results in the field of satellite image processing. However, they require a huge and comprehensive training dataset to avoid overfitting problems and to represent a generalizable model. Thus, moving toward the development of nonsupervised deep learning (NSDL) models in different remote sensing applications is an inevitable need. To provide an initial response to that need, this article performs a comprehensive review and systematic meta-analysis of recently published research articles focusing on the applications of NSDL for remote sensing data processing. In order to identify future research directions and formulate recommendations, we extract trends and highlight interesting approaches from this large body of literature. Consequently, current challenges, prospects, and recommendations are also discussed to uncover the trend. According to the results, there is a sharp increasing trend in the applicability of NSDL methods during these few years particularly, with the advent of new deep architectures, such as adversarial, graph, and transformer models. As a result, this review article discusses different remote sensing data processing applications and challenges that can be addressed using NSDL approaches.

We are witnessing an accelerated technological evolution that has enabled the development of artificial intelligence in various fields, allowing it to gradually infiltrate the entire society. We intend to cover only a small subset of AI technologies in our paper, that of Generative Artificial Intelligence (GenAI). Our objectives are to shed light on the legal issues that GenAI can cause and to find solutions to them. We begin with a definition of GenAI in the much broader context of AI technologies. Answers to a few essential questions are to be found: 'How does GenAI work?', 'What could GenAI be used for?', 'What legal issues could arise from using a GenAI?'. To accomplish our goals, we first conduct a literature review to define artificial intelligence (AI) in general and GenAI in particular. Several lawsuits are chosen to illustrate the magnitude of the legal problems and to test the feasibility of possible solutions in both the national and EU legal systems. Then, we analyse GenAI's output, liability for its contents and for its use, altogether with examples of related contractual clauses.

There has been a significant increase in the availability of global high-resolution land cover (HRLC) datasets due to growing demand and favorable technological advancements. However, this has brought forth the challenge of collecting reference data with a high level of detail for global extents. While photo-interpretation is considered optimal for collecting quality training data for global HRLC mapping, some producers of existing HRLCs use less trustworthy sources, such as existing land cover at a lower resolution, to reduce costs. This work proposes a methodology to extract the most accurate parts of existing HRLCs in response to the challenge of providing reliable reference data at a low cost. The methodology combines existing HRLCs by intersection, and the output represents a Map Of Land Cover Agreement (MOLCA) that can be utilized for selecting training samples. MOLCA's effectiveness was demonstrated through HRLC map production in Africa, in which it generated 48,000 samples. The best classification test had an overall accuracy of 78%. This level of accuracy is comparable to or better than the accuracy of existing HRLCs obtained from more expensive sources of training data, such as photo-interpretation, highlighting the cost-effectiveness and reliability potential of the developed methodology in supporting global HRLC production.

Robotic manipulation requires object pose knowledge for the objects of interest. In order to perform typical household chores, a robot needs to be able to estimate 6D poses for objects such as water glasses or salad bowls. This is especially difficult for glass objects, as for these, depth data are mostly disturbed, and in RGB images, occluded objects are still visible. Thus, in this paper, we propose to redefine the ground-truth for training RGB-based pose estimators in two ways: (a) we apply a transparency-aware multisegmentation, in which an image pixel can belong to more than one object, and (b) we use transparency-aware bounding boxes, which always enclose whole objects, even if parts of an object are formally occluded by another object. The latter approach ensures that the size and scale of an object remain more consistent across different images. We train our pose estimator, which was originally designed for opaque objects, with three different ground-truth types on the ClearPose dataset. Just by changing the training data to our transparency-aware segmentation, with no additional glass-specific feature changes in the estimator, the ADD-S AUC value increases by 4.3%. Such a multisegmentation can be created for every dataset that provides a 3D model of the object and its ground-truth pose. [ABSTRACT FROM AUTHOR]

How to model the effect of reflection is crucial for single image reflection removal (SIRR) task. Modern SIRR methods usually simplify the reflection formulation with the assumption of linear combination of a transmission layer and a reflection layer. However, the large variations in image content and the real-world picture-taking conditions often result in far more complex reflection. In this paper, we introduce a new screen-blur combination based on two important factors, namely the intensity and the blurriness of reflection, to better characterize the reflection formulation in SIRR. Specifically, we present Screen-blur Reflection Networks (SRNet), which executes the screen-blur formulation in its network design and adapts to the complex reflection on real scenes. Technically, SRNet consists of three components: a blended image generator, a reflection estimator and a reflection removal module. The image generator exploits the screen-blur combination to synthesize the training blended images. The reflection estimator learns the reflection layer and a blur degree that measures the level of blurriness for reflection. The reflection removal module further uses the blended image, blur degree and reflection layer to filter out the transmission layer in a cascaded manner. Superior results on three different SIRR methods are reported when generating the training data on the principle of the screen-blur combination. Moreover, extensive experiments on six datasets quantitatively and qualitatively demonstrate the efficacy of SRNet over the state-of-the-art methods. [ABSTRACT FROM AUTHOR]

This paper focuses on the facial micro-expression (FME) generation task, which has potential application in enlarging digital FME datasets, thereby alleviating the lack of training data with labels in existing micro-expression datasets. Despite obvious progress in the image animation task, FME generation remains challenging because existing image animation methods can hardly encode subtle and short-term facial motion information. To this end, we present a facial-prior-guided FME generation framework that takes advantage of facial priors for facial motion generation. Specifically, we first estimate the geometric locations of action units (AUs) with detected facial landmarks. We further calculate an adaptive weighted prior (AWP) map, which alleviates the estimation error of AUs while efficiently capturing subtle facial motion patterns. To achieve smooth and realistic synthesis results, we use our proposed facial prior module to guide motion representation and generation modules in mainstream image animation frameworks. Extensive experiments on three benchmark datasets consistently show that our proposed facial prior module can be adopted in image animation frameworks and significantly improve their performance on micro-expression generation. Moreover, we use the generation technique to enlarge existing datasets, thereby improving the performance of general action recognition backbones on the FME recognition task. Our code is available at https://github.com/sysu19351158/FPB-FOMM. [ABSTRACT FROM AUTHOR]

This research investigates the potential of in vivo learning to enhance visual representation learning for image-based person re-identification (re-ID). Compared to traditional self-supervised learning (which require external data), the introduced in vivo learning utilizes supervisory labels generated from pedestrian images to improve re-ID accuracy without relying on external data sources. Three carefully designed in vivo learning tasks, leveraging statistical regularities within images, are proposed without the need for laborious manual annotations. These tasks enable feature extractors to learn more comprehensive and discriminative person representations by jointly modeling various aspects of human biological structure information, contributing to enhanced re-ID performance. Notably, the method seamlessly integrates with existing re-ID frameworks, requiring minimal modifications and no additional data beyond the existing training set. Extensive experiments on diverse datasets, including Market1501, CUHK03-NP, Celeb-reID, Celeb-reid-light, PRCC, and LTCC, demonstrate substantial enhancements in rank-1 precision compared to state-of-the-art methods. [ABSTRACT FROM AUTHOR]

We are witnessing an accelerated technological evolution that has enabled the development of artificial intelligence in various fields, allowing it to gradually infiltrate the entire society. We intend to cover only a small subset of AI technologies in our paper, that of Generative Artificial Intelligence (GenAI). Our objectives are to shed light on the legal issues that GenAI can cause and to find solutions to them. We begin with a definition of GenAI in the much broader context of AI technologies. Answers to a few essential questions are to be found: 'How does GenAI work?', 'What could GenAI be used for?', 'What legal issues could arise from using a GenAI?'. To accomplish our goals, we first conduct a literature review to define artificial intelligence (AI) in general and GenAI in particular. Several lawsuits are chosen to illustrate the magnitude of the legal problems and to test the feasibility of possible solutions in both the national and EU legal systems. Then, we analyse GenAI's output, liability for its contents and for its use, altogether with examples of related contractual clauses. [ABSTRACT FROM AUTHOR]

The quality of Machine Learning (ML) applications is commonly assessed by quantifying how well an algorithm fits its respective training data. Yet, a perfect model that learns from and reproduces erroneous data will always be flawed in its real-world application. Hence, a comprehensive assessment of ML quality must include an additional data perspective, especially for models trained on human-annotated data. For the collection of human-annotated training data, best practices often do not exist and leave researchers to make arbitrary decisions when collecting annotations. Decisions about the selection of annotators or label options may affect training data quality and model performance. In this paper, I will outline and summarize previous research and approaches to the collection of annotated training data. I look at data annotation and its quality confounders from two perspectives: the set of annotators and the strategy of data collection. The paper will highlight the various implementations of text and image annotation collection and stress the importance of careful task construction. I conclude by illustrating the consequences for future research and applications of data annotation. The paper is intended give readers a starting point on annotated data quality research and stress the necessity of thoughtful consideration of the annotation collection process to researchers and practitioners. [ABSTRACT FROM AUTHOR]

Artificial intelligence systems require large amounts of data to allow them to learn and achieve high performance. That data is increasingly collected in extractive and exploitative ways, which transfer value and power from individuals to AI system owners. Our research focuses on data that is collected from users of digital platforms, through direct and indirect interaction with those platforms, in ways that are not communicated to users, without consent or compensation. This paper presents our findings from a series of interviews and workshops in the Aotearoa New Zealand context to identify common themes and concerns from a variety of perspectives. Reframing this type of interaction as work or labour brings into view an otherwise unrecognised harm of using this data for training AI systems, and illustrates a new class of exploitative data practices that have become normalised in the digital age. We found that participants particularly emphasised moral or ethical justifications for intervention over financial or economic reasons to act. [ABSTRACT FROM AUTHOR]

ABSTRACTArtificial intelligence systems require large amounts of data to allow them to learn and achieve high performance. That data is increasingly collected in extractive and exploitative ways, which transfer value and power from individuals to AI system owners. Our research focuses on data that is collected from users of digital platforms, through direct and indirect interaction with those platforms, in ways that are not communicated to users, without consent or compensation. This paper presents our findings from a series of interviews and workshops in the Aotearoa New Zealand context to identify common themes and concerns from a variety of perspectives. Reframing this type of interaction as work or labour brings into view an otherwise unrecognised harm of using this data for training AI systems, and illustrates a new class of exploitative data practices that have become normalised in the digital age. We found that participants particularly emphasised moral or ethical justifications for intervention over financial or economic reasons to act.

Data-centric artificial intelligence (DC-AI) is a modern paradigm that gives more priority to data quality enhancement, rather than only optimizing the complex codes of AI models. The DC-AI paradigm is expected to substantially advance the status of AI research and developments, which has been solely based on model-centric AI (MC-AI) over the past 30 years. Until present, there exists very little knowledge about DC-AI, and its significance in terms of solving real-world problems remains unexplored in the recent literature. In this technical note, we present the core aspects of DC-AI and MC-AI and discuss their interplay when used to solve some real-world problems. We discuss the potential scenarios/situations that require the integration of DC-AI with MC-AI to solve challenging problems in AI. We performed a case study on a real-world dataset to corroborate the potential of DC-AI in realistic scenarios and to prove its significance over MC-AI when either data are limited or their quality is poor. Afterward, we comprehensively discuss the challenges that currently hinder the realization of DC-AI, and we list promising avenues for future research and development concerning DC-AI. Lastly, we discuss the next-generation computing for DC-AI that can foster DC-AI-related developments and can help transition DC-AI from theory to practice. Our detailed analysis can guide AI practitioners toward exploring the undisclosed potential of DC-AI in the current AI-driven era.

Federated learning (FL) has emerged as one of the de-facto privacy-preserving paradigms that can effectively work with decentralized data sources (e.g., hospitals) without acquiring any private data. Recently, applications of FL have vastly expanded into multiple domains, particularly the medical domain, and FL is becoming one of the mainstream technologies of the near future. In this study, we provide insights into FL fundamental concepts (e.g., the difference from centralized learning, functions of clients and servers, workflows, and nature of data), architecture and applications in the general medical domain, synergies with emerging technologies, key challenges (medical domain), and potential research prospects. We discuss major taxonomies of the FL systems and enlist technical factors in the FL ecosystem that are the foundation of many adversarial attacks on these systems. We also highlight the promising applications of FL in the medical domain by taking the recent COVID-19 pandemic as an application use case. We highlight potential research and development trajectories to further enhance the persuasiveness of this emerging paradigm from the technical point of view. We aim to concisely present the progress of FL up to the present in the medical domain including COVID-19 and to suggest future research trajectories in this area.