Articoli di riviste sul tema "Domain generalisation"
Segui questo link per vedere altri tipi di pubblicazioni sul tema: Domain generalisation.
Cita una fonte nei formati APA, MLA, Chicago, Harvard e in molti altri stili
Vedi i top-50 articoli di riviste per l'attività di ricerca sul tema "Domain generalisation".
Accanto a ogni fonte nell'elenco di riferimenti c'è un pulsante "Aggiungi alla bibliografia". Premilo e genereremo automaticamente la citazione bibliografica dell'opera scelta nello stile citazionale di cui hai bisogno: APA, MLA, Harvard, Chicago, Vancouver ecc.
Puoi anche scaricare il testo completo della pubblicazione scientifica nel formato .pdf e leggere online l'abstract (il sommario) dell'opera se è presente nei metadati.
Vedi gli articoli di riviste di molte aree scientifiche e compila una bibliografia corretta.
1
Zhou, Kaiyang, Yongxin Yang, Timothy Hospedales e Tao Xiang. "Deep Domain-Adversarial Image Generation for Domain Generalisation". Proceedings of the AAAI Conference on Artificial Intelligence 34, n. 07 (3 aprile 2020): 13025–32. http://dx.doi.org/10.1609/aaai.v34i07.7003.
Abstract (sommario):
Machine learning models typically suffer from the domain shift problem when trained on a source dataset and evaluated on a target dataset of different distribution. To overcome this problem, domain generalisation (DG) methods aim to leverage data from multiple source domains so that a trained model can generalise to unseen domains. In this paper, we propose a novel DG approach based on Deep Domain-Adversarial Image Generation (DDAIG). Specifically, DDAIG consists of three components, namely a label classifier, a domain classifier and a domain transformation network (DoTNet). The goal for DoTNet is to map the source training data to unseen domains. This is achieved by having a learning objective formulated to ensure that the generated data can be correctly classified by the label classifier while fooling the domain classifier. By augmenting the source training data with the generated unseen domain data, we can make the label classifier more robust to unknown domain changes. Extensive experiments on four DG datasets demonstrate the effectiveness of our approach.
2
Seemakurthy, Karthik, Charles Fox, Erchan Aptoula e Petra Bosilj. "Domain Generalised Faster R-CNN". Proceedings of the AAAI Conference on Artificial Intelligence 37, n. 2 (26 giugno 2023): 2180–90. http://dx.doi.org/10.1609/aaai.v37i2.25312.
Abstract (sommario):
Domain generalisation (i.e. out-of-distribution generalisation) is an open problem in machine learning, where the goal is to train a model via one or more source domains, that will generalise well to unknown target domains. While the topic is attracting increasing interest, it has not been studied in detail in the context of object detection. The established approaches all operate under the covariate shift assumption, where the conditional distributions are assumed to be approximately equal across source domains. This is the first paper to address domain generalisation in the context of object detection, with a rigorous mathematical analysis of domain shift, without the covariate shift assumption. We focus on improving the generalisation ability of object detection by proposing new regularisation terms to address the domain shift that arises due to both classification and bounding box regression. Also, we include an additional consistency regularisation term to align the local and global level predictions. The proposed approach is implemented as a Domain Generalised Faster R-CNN and evaluated using four object detection datasets which provide domain metadata (GWHD, Cityscapes, BDD100K, Sim10K) where it exhibits a consistent performance improvement over the baselines. All the codes for replicating the results in this paper can be found at https://github.com/karthikiitm87/domain-generalisation.git
3
Le, Hoang Son, Rini Akmeliawati e Gustavo Carneiro. "Domain Generalisation with Domain Augmented Supervised Contrastive Learning (Student Abstract)". Proceedings of the AAAI Conference on Artificial Intelligence 35, n. 18 (18 maggio 2021): 15821–22. http://dx.doi.org/10.1609/aaai.v35i18.17907.
Abstract (sommario):
Domain generalisation (DG) methods address the problem of domain shift, when there is a mismatch between the distributions of training and target domains. Data augmentation approaches have emerged as a promising alternative for DG. However, data augmentation alone is not sufficient to achieve lower generalisation errors. This project proposes a new method that combines data augmentation and domain distance minimisation to address the problems associated with data augmentation and provide a guarantee on the learning performance, under an existing framework. Empirically, our method outperforms baseline results on DG benchmarks.
4
Roeper, P. "Generalisation of first-order logic to nonatomic domains". Journal of Symbolic Logic 50, n. 3 (settembre 1985): 815–38. http://dx.doi.org/10.2307/2274334.
Abstract (sommario):
The quantifiers of standard predicate logic are interpreted as ranging over domains of individuals, and interpreted formulae beginning with a quantifier make claims to the effect that something is true of every individual, i.e. of the whole domain, or of some individuals, i.e. of part of the domain. To state that something is true of all or part of a totality seems to be the basic significance of universal and existential quantification, and this by itself does not involve a specification of the structure of the totality. This means that the notion of quantification by itself does not demand totalities of individuals, i.e. atomic totalities, as domains of quantification. Nonatomic domains, such as volumes of space, or surfaces, are equally in order. So one might say that a certain predicate applies “everywhere” or “somewhere” in such a domain. All that the concept of quantification requires is a totality which is structured in terms of a part-to-whole relation, and appropriate properties that apply to part or all of the totality. Quantification does not demand that the totality have smallest parts, or atoms. There is no conflict with the sense of universal or existential quantification if the domain is nonatomic, if every one of its parts has itself proper parts.The most general kind of quantification theory must then deal with totalities of any kind, atomic or not. The relationships among the parts of a domain are described by the theory of Boolean algebras, which we can regard as the most general characterisation of a totality, of a domain of quantification.In this paper I shall be concerned with this generalised theory of quantification, which encompasses nonatomic domains as well as atomic and mixed domains, i.e. totalities consisting entirely or partly of individuals.
5
Gomathi, R., e S. Selvakumaran. "A Novel Medical Image Segmentation Model with Domain Generalization Approach". International Journal of Electrical and Electronics Research 10, n. 2 (30 giugno 2022): 312–19. http://dx.doi.org/10.37391/ijeer.100242.
Abstract (sommario):
In deep learning-based computing vision for image processing, image segmentation is a prominent issue. There is promising generalisation performance in the medical image segmentation sector for approaches using domain generalisation (DG). Single domain generalisation (SDG) is a more difficult problem than conventional generalisation (DG), which requires numerous source domains to be accessible during network training, as opposed to conventional generalisation (DG). Color medical images may be incorrectly segmented because of the augmentation of the full image in order to increase model generalisation capacity. An arbitrary illumination SDG model for improving generalisation power for colour image segmentation approach for medical images through synthesizing random radiance charts is presented as a first solution to this challenge. Color medical images may be decomposed into reflectivity and illumination maps using retinex-based neural networks (ID-Nets). In order to provide medical colour images under various lighting situations, illumination randomization is used to enhance illumination maps. A new metric, TGCI, called the transfer gradient consistency index was devised to quantify the performance of the breakdown of retinal images by simulating physical lighting. Two of the existing retinal image segmentation tasks are tested extensively in order to assess our suggested system. According to the Dice coefficient, our framework surpasses previous SDGs and image improvement algorithms, outperforming the best SDGs by up to 1.7 per cent.
6
Myers, Scott, e Jaye Padgett. "Domain generalisation in artificial language learning". Phonology 31, n. 3 (dicembre 2014): 399–433. http://dx.doi.org/10.1017/s0952675714000207.
Abstract (sommario):
Many languages have restrictions on word-final segments, such as a requirement that any word-final obstruent be voiceless. There is a phonetic basis for such restrictions at the ends of utterances, but not the ends of words. Historical linguists have long noted this mismatch, and have attributed it to an analogical generalisation of such restrictions from utterance-final to word-final position. To test whether language learners actually generalise in this way, two artificial language learning experiments were conducted. Participants heard nonsense utterances in which there was a restriction on utterance-final obstruents, but in which no information was available about word-final utterance-medial obstruents. They were then tested on utterances that included obstruents in both positions. They learned the pattern and generalised it to word-final utterance-medial position, confirming that learners are biased toward word-based distributional patterns.
7
HYLAND, MARTIN. "Some reasons for generalising domain theory". Mathematical Structures in Computer Science 20, n. 2 (25 marzo 2010): 239–65. http://dx.doi.org/10.1017/s0960129509990375.
Abstract (sommario):
One natural way to generalise domain theory is to replace partially ordered sets by categories. This kind of generalisation has recently found application in the study of concurrency. An outline is given of the elegant mathematical foundations that have been developed. This is specialised to give a construction of cartesian closed categories of domains, which throws light on standard presentations of domain theory.
8
Rashid, Fariza, Ben Doyle, Soyeon Caren Han e Suranga Seneviratne. "Phishing URL detection generalisation using Unsupervised Domain Adaptation". Computer Networks 245 (maggio 2024): 110398. http://dx.doi.org/10.1016/j.comnet.2024.110398.
9
Anthony, Martin, Peter Bartlett, Yuval Ishai e John Shawe-Taylor. "Valid Generalisation from Approximate Interpolation". Combinatorics, Probability and Computing 5, n. 3 (settembre 1996): 191–214. http://dx.doi.org/10.1017/s096354830000198x.
Abstract (sommario):
Let and be sets of functions from domain X to ℝ. We say that validly generalises from approximate interpolation if and only if for each η > 0 and ∈, δ ∈ (0,1) there is m0(η, ∈, δ) such that for any function t ∈ and any probability distribution on X, if m > m0 then with m-probability at least 1 – δ, a sample X = (x1, X2,…,xm) ∈ Xm satisfiesWe find conditions that are necessary and sufficient for to validly generalise from approximate interpolation, and we obtain bounds on the sample length m0{η,∈,δ) in terms of various parameters describing the expressive power of .
10
Orazov, B. B. "On the asymptotic behaviour at infinity of solutions of the traction boundary value problem". Proceedings of the Royal Society of Edinburgh: Section A Mathematics 111, n. 1-2 (1989): 33–52. http://dx.doi.org/10.1017/s0308210500024999.
Abstract (sommario):
SynopsisKorn's inequalities are proved for star-shaped domains and it is shown how the constants in these inequalities depend on the dimensions of the domain. These inequalities are then used to prove a generalisation of Saint-Venant's Principle for nonlinear elasticity and additionally to establish the asymptotic behaviour of solutions to the traction boundary value problem for a non-prismatic cylinder.
11
Bucur, Dorin, e Paola Trebeschi. "Shape optimisation problems governed by nonlinear state equations". Proceedings of the Royal Society of Edinburgh: Section A Mathematics 128, n. 5 (1998): 945–63. http://dx.doi.org/10.1017/s0308210500030006.
Abstract (sommario):
The purpose of this paper is to give a compactness-continuity result for the solution of a nonlinear Dirichlet problem in terms of its domain variation. The topology in the family of domains is given by the Hausdorff metric and continuity is obtained under capacity conditions. A generalisation of Sverak's result in iV-dimensions is deduced as a particular case.
12
Aho, Kaarina, Brett D. Roads e Bradley C. Love. "System alignment supports cross-domain learning and zero-shot generalisation". Cognition 227 (ottobre 2022): 105200. http://dx.doi.org/10.1016/j.cognition.2022.105200.
13
Lennings, C. J. "Adolescent Time Perspective: A Further Note". Perceptual and Motor Skills 77, n. 3 (dicembre 1993): 808–10. http://dx.doi.org/10.2466/pms.1993.77.3.808.
Abstract (sommario):
Researchers on temporal extension have moved from using global measures of temporal extension to using domain-specific measures. A reanalysis of 1992 data from 98 boys and 103 girls tested whether domain-specific measures of temporal extension were correlated with global measures of both past and future temporal extension. Moderate correlations support the generalisation of findings from earlier studies, using global measures, to more recent studies with domain-specific measures.
14
ZHANG, LI, WEIDA ZHOU e LICHENG JIAO. "SUPPORT VECTOR MACHINES BASED ON THE ORTHOGONAL PROJECTION KERNEL OF FATHER WAVELET". International Journal of Computational Intelligence and Applications 05, n. 03 (settembre 2005): 283–303. http://dx.doi.org/10.1142/s1469026805001489.
Abstract (sommario):
Recently the study on the theory of wavelets shows that the wavelets have not only the multi-resolution property both in frequency and time domain, but also the good approximation ability. SVMs based on the statistical learning theory are a kind of general and effective learning machines, and have described for us the nice application blueprint in machine learning domain. There exists a bottleneck problem, or the pre-selection of kernel parameter for SVMs. In this paper, the orthogonal projection kernels of father wavelet (OPFW kernels) are introduced into SVMs. In doing so SVMs based on the OPFW kernels can have good performance in both approximation and generalisation. Simultaneously the parameter pre-selection of wavelet kernels can be implemented by discrete wavelet transform. Experiments on regression estimation illustrate the approximation and generalisation ability of our method.
15
Guo, Zongming, e J. R. L. Webb. "Uniqueness of positive solutions for quasilinear elliptic equations when a parameter is large". Proceedings of the Royal Society of Edinburgh: Section A Mathematics 124, n. 1 (1994): 189–98. http://dx.doi.org/10.1017/s0308210500029280.
Abstract (sommario):
Existence and uniqueness results are proved for positive solutions of a class of quasilinear elliptic equations in a domain Ω⊂ℝN via a generalisation of Serrin's sweeping principle. In the case when Ω is an annulus, it is shown that the solution is radially symmetric.
16
Edgar, Charlotte, Faye Chan, Thomas Armstrong e Edwin S. Dalmaijer. "Long-term disgust habituation with limited generalisation in care home workers". PLOS ONE 19, n. 4 (17 aprile 2024): e0299429. http://dx.doi.org/10.1371/journal.pone.0299429.
Abstract (sommario):
Countless workers handle bodily effluvia and body envelope violations every working day, and consequentially face deeply unpleasant levels of disgust. Understanding if and how they adapt can help inform policies to improve worker satisfaction and reduce staff turnover. So far, limited evidence exist that self-reported disgust is reduced (or lower to begin with) among those employed in high-disgust environments. However, it is unclear if this is due to demand effects or translates into real behavioural changes. Here, we tested healthcare assistants (N = 32) employed in UK care homes and a control sample (N = 50). We replicated reduced self-reported pathogen disgust sensitivity in healthcare workers compared to controls. We also found it negatively correlated with career duration, suggesting long-term habituation. Furthermore, we found that healthcare assistants showed no behavioural disgust avoidance on a web-based preferential looking task (equivalent to eye tracking). Surprisingly, this extended to disgust elicitors found outside care homes, suggesting generalisation of disgust habituation. While we found no difference between bodily effluvia (core disgust) and body envelope violations (gore disgust), generalisation did not extend to other domains: self-reported sexual and moral disgust sensitivity were not different between healthcare assistants and the control group, nor was there a correlation with career duration. In sum, our work confirms that people in high-frequency disgust employment are less sensitive to pathogen disgust. Crucially, we provide preliminary evidence that this is due to a process of long-term habituation with generalisation to disgust-elicitors within the pathogen domain, but not beyond it.
17
Moretti, N., X. Xie, J. Merino Garcia, J. Chang e A. Kumar Parlikad. "Digital Twin based built environment asset management services development". IOP Conference Series: Earth and Environmental Science 1101, n. 9 (1 novembre 2022): 092023. http://dx.doi.org/10.1088/1755-1315/1101/9/092023.
Abstract (sommario):
Abstract The Digital Twin (DT) concept emerged recently in the Architecture, Engineering, Constructions and Operations domain and the interest for digital built asset management service implementations is increasing. However, most of the applications are developed according to custom approaches, preventing the generalisation and modularisation. This article presents a methodological framework for DT-based applications development. DT-based applications support effective data-centric processes in the management of critical assets. The proposed approach is based on a flexible and federated data model, enabling dynamic data management. The federated data model provides an interface and connection among different knowledge domains. The service requirement definition provides the input specification for the application development, accomplished through the modelling of three fundamental components: data, processes, and algorithms. These components are assembled through Application Programming Interfaces (APIs), supporting the DT-based services. The proposed approach is applied to the West Cambridge DT research facility and tested in the building energy sub-metering strategy. The developed approach allows flexible and interoperable data-centric applications development, contributing to the generalisation and scalability of DT-based built asset management applications. This aims at providing a consistent approach that can be employed in different domains, where the DT technologies play a central role in service innovation.
18
Liu, Yifan, Tiecheng Song, Chengye Xian, Ruiyuan Chen, Yi Zhao, Rui Li e Tan Guo. "Two-Stage Adaptive Network for Semi-Supervised Cross-Domain Crater Detection under Varying Scenario Distributions". Remote Sensing 16, n. 11 (5 giugno 2024): 2024. http://dx.doi.org/10.3390/rs16112024.
Abstract (sommario):
Crater detection can provide valuable information for humans to explore the topography and understand the history of extraterrestrial planets. Due to the significantly varying scenario distributions, existing detection models trained on known labelled crater datasets are hardly effective when applied to new unlabelled planets. To address this issue, we propose a two-stage adaptive network (TAN) for semi-supervised cross-domain crater detection. Our network is built on the YOLOv5 detector, where a series of strategies are employed to enhance its cross-domain generalisation ability. In the first stage, we propose an attention-based scale-adaptive fusion (ASAF) strategy to handle objects with significant scale variances. Furthermore, we propose a smoothing hard example mining (SHEM) loss function to address the issue of overfitting on hard examples. In the second stage, we propose a sort-based pseudo-labelling fine-tuning (SPF) strategy for semi-supervised learning to mitigate the distributional differences between source and target domains. For both stages, we employ weak or strong image augmentation to suit different cross-domain tasks. Experimental results on benchmark datasets demonstrate that the proposed network can enhance domain adaptation ability for crater detection under varying scenario distributions.
19
Tulchinskii, G. L. "Argumentum ad morti in the violence discourse: the semantics and pragmatics of ‘radical’ argumentation". Slovo.ru: Baltic accent 11, n. 4 (2020): 58–64. http://dx.doi.org/10.5922/2225-5346-2020-4-5.
Abstract (sommario):
The appeal to death is a type of argument that either explicitly or implicitly invokes human finitude. This rhetorical device contributes to the credibility of requests, wishes, etc., or blocks communication. The illocutionary power of the appeal to death is determined by the means it uses, which range from inductive generalisation and deductions to approaching the limit of this generalisation, to the threat of going beyond the existential boundaries of the discourse, to the problem domain losing its existential status, and to the threat being carried out. All of them make communication insignificant and the subject, or even the opponent, non-existent. This style of argumentation is peculiar to ‘pre-logical’ communities where the logical culture of social communication is underdeveloped and unwanted, where an instrumental appeal is made not to the law but force, not to freedom but arbitrariness.
20
Maia, Macedo, e Markus Endres. "CQA Transformer Models in the Home Improvement Domain". Journal of Data Intelligence 3, n. 1 (febbraio 2022): 131–48. http://dx.doi.org/10.26421/jdi3.1-3.
Abstract (sommario):
To find answers for subjective questions about many topics through Q\&A forums, questioners and answerers can cooperatively help themselves by sharing their doubts or answers based on their background and life experiences. These experiences can help machines redirect questioners to find better answers based on community question-answering models. This work proposes a comparative analysis of the pairwise community answer retrieval models in the home improvement domain considering different kinds of user question context information. Community Question-Answering (CQA) models must rank candidate answers in decreasing order of relevance for a user question. Our contribution consists of transformer-based language models using different kinds of user information to accurate the model generalisation. To train our model, we propose a proper CQA dataset in the home improvement domain that consists of information extracted from community forums, including question context information. We evaluate our approach by comparing the performance of each baseline model based on rank-aware evaluation measures.
21
Song, Yunlin, Hua Zheng, Xiaoping Lu e Seak-Weng Vong. "A Two-Step Iteration Method for Vertical Linear Complementarity Problems". Symmetry 14, n. 9 (8 settembre 2022): 1882. http://dx.doi.org/10.3390/sym14091882.
Abstract (sommario):
In this paper, a two-step iteration method was established which can be viewed as a generalisation of the existing modulus-based methods for vertical linear complementarity problems. The convergence analysis of the proposed method is presented, which can enlarge the convergence domain of the parameter matrix compared to the recent results. Numerical examples show that the proposed method is efficient with the two-step technique and confirm the improvement of the theoretical results.
22
Verenich, Vadim. "Neural Networks in Legal Theory". Studia Humana 13, n. 3 (28 aprile 2024): 41–51. http://dx.doi.org/10.2478/sh-2024-0018.
Abstract (sommario):
Abstract This article explores the domain of legal analysis and its methodologies, emphasising the significance of generalisation in legal systems. It discusses the process of generalisation in relation to legal concepts and the development of ideal concepts that form the foundation of law. The article examines the role of logical induction and its similarities with semantic generalisation, highlighting their importance in legal decision-making. It also critiques the formal-deductive approach in legal practice and advocates for more adaptable models, incorporating fuzzy logic, non-monotonic defeasible reasoning, and artificial intelligence. The potential application of neural networks, specifically deep learning algorithms, in legal theory is also discussed. The article discusses how neural networks encode legal knowledge in their synaptic connections, while the syllogistic model condenses legal information into axioms. The article also highlights how neural networks assimilate novel experiences and exhibit evolutionary progression, unlike the deductive model of law. Additionally, the article examines the historical and theoretical foundations of jurisprudence that align with the basic principles of neural networks. It delves into the statistical analysis of legal phenomena and theories that view legal development as an evolutionary process. The article then explores Friedrich Hayek’s theory of law as an autonomous self-organising system and its compatibility with neural network models. It concludes by discussing the implications of Hayek’s theory on the role of a lawyer and the precision of neural networks.
23
Hill, Adrian T., e Endre Süli. "Dynamics of a nonlinear convection-diffusion equation in multidimensional bounded domains". Proceedings of the Royal Society of Edinburgh: Section A Mathematics 125, n. 2 (1995): 439–48. http://dx.doi.org/10.1017/s0308210500028110.
Abstract (sommario):
The scalar nonlinear convection-diffusion equationis considered, for given initial data and zero Dirichlet boundary conditions, in a smooth bounded domain Ω⊂ℝn. The homogeneous viscous Burgers' equation in one dimension is well-known to possess a unique, exponentially attracting equilibrium. These properties are shown to be preserved in the generalisation considered. Furthermore, the equilibrium is shown to be bounded in the maximum norm independently of the function a. The main methods used are maximum principles, and a variational method due to Stampacchia.
24
Kolberg, Jascha, Marta Gomez‐Barrero e Christoph Busch. "On the generalisation capabilities of fingerprint presentation attack detection methods in the short wave infrared domain". IET Biometrics 10, n. 4 (9 marzo 2021): 359–73. http://dx.doi.org/10.1049/bme2.12020.
25
Amieur, Khalissa Derbal, Kamel Boukhalfa e Zaia Alimazighi. "Geo-Multi-Agent System Based Webmapping Approach Using Multiple Representation and Generalisation Driven by Domain Ontology". International Journal of Information Technology and Web Engineering 8, n. 4 (ottobre 2013): 48–64. http://dx.doi.org/10.4018/ijitwe.2013100104.
Abstract (sommario):
Geographic Information (GI) is currently available at any time, from anywhere on the surface of the earth, for any person connected to internet. Some applications of design, implementation, generation and dissemination of maps on the web are recognized as “Webmapping” application, geographic web services or more generally on demand-map making tools. All these web applications aims the satisfaction of user needs by providing personalized maps in a fast response time with a good quality. However, the complexity and diversity of aspects taking into account have lead researchers to focus on one aspect at the expense of others. Consequently, few works have addressed all these aspects simultaneously. The authors propose in this paper, a Webmapping approach organized into two main tasks: (1) query analysis driven by domain ontology in analyzing a query launched by a user on a web browser and (2) map generation process. The first step allows extracting and formalizing user needs through two map determinants factors: the Level of Detail (LoD) and Point of View (PoV) and the second, exploit an hybrid approach “Multi Representation and Generalization” in storing and generating geographical data with integrating Multi-Agent technology in all steps of processing. To evaluate the effectiveness of our proposal, a first tool prototype implementing our approach is so developed using a geographic vector dataset provided by national cartographic agency.
26
Wu, Guile, e Shaogang Gong. "Generalising without Forgetting for Lifelong Person Re-Identification". Proceedings of the AAAI Conference on Artificial Intelligence 35, n. 4 (18 maggio 2021): 2889–97. http://dx.doi.org/10.1609/aaai.v35i4.16395.
Abstract (sommario):
Existing person re-identification (Re-ID) methods mostly prepare all training data in advance, while real-world Re-ID data are inherently captured over time or from different locations, which requires a model to be incrementally generalised from sequential learning of piecemeal new data without forgetting what is already learned. In this work, we call this lifelong person Re-ID, characterised by solving a problem of unseen class identification subject to continuous new domain generalisation and adaptation with class imbalanced learning. We formulate a new Generalising without Forgetting method (GwFReID) for lifelong Re-ID and design a comprehensive learning objective that accounts for classification coherence, distribution coherence and representation coherence in a unified framework. This design helps to simultaneously learn new information, distil old knowledge and solve class imbalance, which enables GwFReID to incrementally improve model generalisation without catastrophic forgetting of what is already learned. Extensive experiments on eight Re-ID benchmarks, CIFAR-100 and ImageNet show the superiority of GwFReID over the state-of-the-art methods.
27
VYTINIOTIS, DIMITRIOS, SIMON PEYTON JONES, TOM SCHRIJVERS e MARTIN SULZMANN. "OutsideIn(X) Modular type inference with local assumptions". Journal of Functional Programming 21, n. 4-5 (16 maggio 2011): 333–412. http://dx.doi.org/10.1017/s0956796811000098.
Abstract (sommario):
AbstractAdvanced type system features, such as GADTs, type classes and type families, have proven to be invaluable language extensions for ensuring data invariants and program correctness. Unfortunately, they pose a tough problem for type inference when they are used as local type assumptions. Local type assumptions often result in the lack of principal types and cast the generalisation of local let-bindings prohibitively difficult to implement and specify. User-declared axioms only make this situation worse. In this paper, we explain the problems and – perhaps controversially – argue for abandoning local let-binding generalisation. We give empirical results that local let generalisation is only sporadically used by Haskell programmers. Moving on, we present a novel constraint-based type inference approach for local type assumptions. Our system, called OutsideIn(X), is parameterised over the particular underlying constraint domain X, in the same way as HM(X). This stratification allows us to use a common metatheory and inference algorithm. OutsideIn(X) extends the constraints of X by introducing implication constraints on top. We describe the strategy for solving these implication constraints, which, in turn, relies on a constraint solver for X. We characterise the properties of the constraint solver for X so that the resulting algorithm only accepts programs with principal types, even when the type system specification accepts programs that do not enjoy principal types. Going beyond the general framework, we give a particular constraint solver for X = type classes + GADTs + type families, a non-trivial challenge in its own right. This constraint solver has been implemented and distributed as part of GHC 7.
28
Heinrich, Mattias P., Hanna Siebert, Laura Graf, Sven Mischkewitz e Lasse Hansen. "Robust and Realtime Large Deformation Ultrasound Registration Using End-to-End Differentiable Displacement Optimisation". Sensors 23, n. 6 (7 marzo 2023): 2876. http://dx.doi.org/10.3390/s23062876.
Abstract (sommario):
Image registration for temporal ultrasound sequences can be very beneficial for image-guided diagnostics and interventions. Cooperative human–machine systems that enable seamless assistance for both inexperienced and expert users during ultrasound examinations rely on robust, realtime motion estimation. Yet rapid and irregular motion patterns, varying image contrast and domain shifts in imaging devices pose a severe challenge to conventional realtime registration approaches. While learning-based registration networks have the promise of abstracting relevant features and delivering very fast inference times, they come at the potential risk of limited generalisation and robustness for unseen data; in particular, when trained with limited supervision. In this work, we demonstrate that these issues can be overcome by using end-to-end differentiable displacement optimisation. Our method involves a trainable feature backbone, a correlation layer that evaluates a large range of displacement options simultaneously and a differentiable regularisation module that ensures smooth and plausible deformation. In extensive experiments on public and private ultrasound datasets with very sparse ground truth annotation the method showed better generalisation abilities and overall accuracy than a VoxelMorph network with the same feature backbone, while being two times faster at inference.
29
Badawi, Ayman. "On 2-absorbing ideals of commutative rings". Bulletin of the Australian Mathematical Society 75, n. 3 (giugno 2007): 417–29. http://dx.doi.org/10.1017/s0004972700039344.
Abstract (sommario):
Suppose that R is a commutative ring with 1 ≠ 0. In this paper, we introduce the concept of 2-absorbing ideal which is a generalisation of prime ideal. A nonzero proper ideal I of R is called a 2-absorbing ideal of R if whenever a, b, c ∈ R and abc ∈ I, then ab ∈ I or ac ∈ I or bc ∈ I. It is shown that a nonzero proper ideal I of R is a 2-absorbing ideal if and only if whenever I1I2I3 ⊆ I for some ideals I1,I2,I3 of R, then I1I2 ⊆ I or I2I3 ⊆ I or I1I3 ⊆ I. It is shown that if I is a 2-absorbing ideal of R, then either Rad(I) is a prime ideal of R or Rad(I) = P1 ⋂ P2 where P1,P2 are the only distinct prime ideals of R that are minimal over I. Rings with the property that every nonzero proper ideal is a 2-absorbing ideal are characterised. All 2-absorbing ideals of valuation domains and Prüfer domains are completely described. It is shown that a Noetherian domain R is a Dedekind domain if and only if a 2-absorbing ideal of R is either a maximal ideal of R or M2 for some maximal ideal M of R or M1M2 where M1,M2 are some maximal ideals of R. If RM is Noetherian for each maximal ideal M of R, then it is shown that an integral domain R is an almost Dedekind domain if and only if a 2-absorbing ideal of R is either a maximal ideal of R or M2 for some maximal ideal M of R or M1M2 where M1,M2 are some maximal ideals of R.
30
Evans, M. W. "Torsion theories over semihereditary rings". Journal of the Australian Mathematical Society. Series A. Pure Mathematics and Statistics 40, n. 1 (febbraio 1986): 54–70. http://dx.doi.org/10.1017/s1446788700026495.
Abstract (sommario):
AbstractIn this paper the class of rings for which the right flat modules form the torsion-free class of a hereditary torsion theory (G, ℱ) are characterized and their structure investigated. These rings are called extended semihereditary rings. It is shown that the class of regular rings with ring homomorphism is a full co-reflective subcategory of the class of extended semihereditary rings with “flat” homomorphisms. A class of prime torsion theories is introduced which determines the torsion theory (G, ℱG). The torsion theory (JG, ℱG) is used to find a suitable generalisation of Dedekind Domain.
31
Dhouib, H., V. Prat, T. Van Reeth e S. Mathis. "The traditional approximation of rotation for rapidly rotating stars and planets". Astronomy & Astrophysics 652 (agosto 2021): A154. http://dx.doi.org/10.1051/0004-6361/202140615.
Abstract (sommario):
Context. The traditional approximation of rotation (TAR) is a treatment of the hydrodynamic equations of rotating and stably stratified fluids in which the action of the Coriolis acceleration along the direction of the entropy and chemical stratifications is neglected because it is weak in comparison with the buoyancy Archimedean force. This leads to the neglect of the horizontal projection of the rotation vector in the equations for the dynamics of gravito-inertial waves (GIWs). The dependent variables in those equations then become separable into radial and horizontal parts as in the non-rotating case. The TAR is built on the assumptions that the star is spherical (i.e., its centrifugal deformation is neglected) and uniformly rotating. However, it has recently been generalised to include the effects of a moderate centrifugal deformation using a perturbative approach. Aims. We study the feasibility of carrying out a new generalisation to account for the centrifugal acceleration in the case of strongly deformed uniformly and rapidly rotating stars (and planets), and to identify the validity domain of this approximation. Methods. We built a complete formalism analytically that allows the study of the dynamics of GIWs in spheroidal coordinates which take the flattening of uniformly and rapidly rotating stars into account by assuming the hierarchies of frequencies adopted within the TAR in the spherical case. Results. Using 2D stellar models, we determine the validity domain of the generalised TAR as a function of the rotation rate of the star normalised by its critical angular velocity and its pseudo-radius. Assuming the anelastic and the two-dimensional Jeffreys-Wentzel-Kramers-Brillouin approximations, we derive a generalised Laplace tidal equation for the horizontal eigenfunctions of the GIWs and their asymptotic wave periods, which can be used to probe the structure and dynamics of rotating deformed stars with asteroseismology. The generalised TAR where the centrifugal deformation of a star (or planet) is taken into account non-perturbatively allows us to identify, within the framework of 2D Evolution STEllaire en Rotation models, the validity domain of this approximation which is reduced by increasing the rate of rotation. We can affirm with a level of confidence of 90% that the TAR remains applicable in all the space domain of deformed stars rotating at a rotation rate lower than 20% of the critical rotation rate. Conclusions. A new generalisation of the TAR, which takes the centrifugal acceleration into account in a non-perturbative way, is derived. This generalisation allows us to study the detectability and the signature of the centrifugal effects on GIWs in rapidly rotating deformed stars (and planets). We found that the effects of the centrifugal acceleration in rapidly rotating early-type stars on GIWs are theoretically detectable in modern space photometry using observations from Kepler. We found also, by comparing the period spacing pattern computed with the standard and the generalised TAR, that the centrifugal acceleration affects the period spacing by increasing its values for low radial orders and by decreasing them slightly for high radial orders.
32
Tu, FengMiao, MingHui Wei, Jun Liu, LiXia Jiang e Jia Zhang. "Metal-loss defect depth inversion in oil and gas pipelines based on Bayesian regularisation neural network". Insight - Non-Destructive Testing and Condition Monitoring 64, n. 12 (1 dicembre 2022): 680–87. http://dx.doi.org/10.1784/insi.2022.64.12.680.
Abstract (sommario):
Defect depth inversion is generally considered as a challenge in magnetic flux leakage (MFL) testing and evaluation because of its strong non-linearity and low prediction accuracy. Current inversion models focus on the inversion accuracy of specific datasets, ignoring consideration of the generalisation ability of inversion models under different conditions. In order to solve such problems, this paper proposes a novel pipeline defect inversion method based on a Bayesian regularisation neural network (BRNN) model. This method consists of two parts. Firstly, three domain features are extracted and a Boruta algorithm is introduced to reduce the feature dimension and obtain the best feature subset. Secondly, in order to approximate the complex non-linear relationship between multi-dimensional features and defect depth, a back-propagation neural network (BPNN) model based on Levenberg-Marquardt optimisation and a Bayesian learning algorithm is constructed. The model can effectively find a close global minimum and overcome the phenomena of overfitting and overtraining. In order to evaluate the performance of the proposed defect inversion method, a comparative experiment is carried out with other well-known inversion algorithms. The results obtained confirm that the inversion method can improve the prediction accuracy of defect depth. More importantly, this method enhances the generalisation ability of defect inversion problems with different sample sets.
33
Husain, Shabbir RV. "Sales Force Structure: A Review". Marketing Review 18, n. 3 (31 dicembre 2018): 227–42. http://dx.doi.org/10.1362/146934718x15434305916826.
Abstract (sommario):
This paper summarises significant developments in research in the sales force structure domain for the facilitation of impactful problem-driven research in the future. The decision-making choices in sales force structuring from the perspective of the salesperson are reviewed, and also specialisation, coordination and control. The impact of stage of the firm and its geographical/cultural context on sales force structure is also explored. The paper culminates with hypotheses around the impact of sales force structuring choices with respect to specialisation, generalisation, formalisation and control on organisational performance, after incorporating the moderating elements related to the growth stage of the firm and the geographical/cultural context.
34
Romani, Giulio. "A-priori bounds for quasilinear problems in critical dimension". Advances in Nonlinear Analysis 9, n. 1 (20 settembre 2019): 788–802. http://dx.doi.org/10.1515/anona-2020-0025.
Abstract (sommario):
Abstract We establish uniform a-priori bounds for solutions of the quasilinear problems $$\begin{array}{} \displaystyle \begin{cases} -{\it\Delta}_Nu=f(u)\quad&\mbox{in }{\it\Omega},\\ u=0\quad&\mbox{on }{\partial{\it\Omega}}, \end{cases} \end{array}$$ where Ω ⊂ ℝN is a bounded smooth convex domain and f is positive, superlinear and subcritical in the sense of the Trudinger-Moser inequality. The typical growth of f is thus exponential. Finally, a generalisation of the result for nonhomogeneous nonlinearities is given. Using a blow-up approach, this paper completes the results in [1, 2], extending the class of nonlinearities for which the uniform a-priori bound applies.
35
Dhouib, H., V. Prat, T. Van Reeth e S. Mathis. "The traditional approximation of rotation for rapidly rotating stars and planets". Astronomy & Astrophysics 656 (dicembre 2021): A122. http://dx.doi.org/10.1051/0004-6361/202141152.
Abstract (sommario):
Context. We examine the dynamics of low-frequency gravito-inertial waves (GIWs) in differentially rotating deformed radiation zones in stars and planets by generalising the traditional approximation of rotation (TAR). The TAR treatment was built on the assumptions that the star is spherical (i.e. its centrifugal deformation is neglected) and uniformly rotating. However, it has been generalised in our previous work by including the effects of the centrifugal deformation using a non-perturbative approach. In the meantime, TAR has been generalised in spherical geometry to take the differential rotation into account. Aims. We aim to carry out a new generalisation of the TAR treatment to account for the differential rotation and the strong centrifugal deformation simultaneously. Methods. We generalise our previous work by taking into account the differential rotation in the derivation of our complete analytical formalism that allows the study of the dynamics of GIWs in differentially and rapidly rotating stars. Results. We derived the complete set of equations that generalises the TAR, simultaneously taking the full centrifugal acceleration and the differential rotation into account. Within the validity domain of the TAR, we derived a generalised Laplace tidal equation for the horizontal eigenfunctions and asymptotic wave periods of the GIWs, which can be used to probe the structure and dynamics of differentially rotating deformed stars with asteroseismology. Conclusions. A new generalisation of the TAR, which simultaneously takes into account the differential rotation and the centrifugal acceleration in a non-perturbative way, was derived. This generalisation allowed us to study the detectability and the signature of the differential rotation on GIWs in rapidly rotating deformed stars and planets. We found that the effects of the differential rotation in early-type deformed stars on GIWs is theoretically largely detectable in modern space photometry using observations from Kepler and TESS.
36
Das, Arghya, e Mustansir Barma. "Fluctuation dominated phase ordering in coarse-grained depth models: domain wall structures, extreme values and coarsening". Journal of Statistical Mechanics: Theory and Experiment 2023, n. 4 (1 aprile 2023): 043203. http://dx.doi.org/10.1088/1742-5468/acc729.
Abstract (sommario):
Abstract Models of particles driven by a one-dimensional fluctuating surface are known to exhibit fluctuation dominated phase ordering (FDPO), in which both the order and fluctuations appear on macroscopic scales. Highly dynamic and macroscopically broad interfacial regions, each composed of many domain walls, appear between macroscopically ordered regions and consequently the scaled correlation function violates the Porod law. We focus on two essential quantities which together quantify the unique characteristics of FDPO, namely the total number of domain walls and the length of the largest ordered domain. We present results in the context of coarse-grained depth (CD) models, both in steady state and while coarsening. Analytic arguments supported by numerical simulations show that even though domain wall number fluctuations are very strong, the associated variance remains constant in time during coarsening. Further, the length of the largest cluster grows as a power law with significant multiplicative logarithms which involve both the time and system size. In addition, we identify corrections to the leading power law scaling in several quantities in the coarsening regime. We also study a generalisation of the CD model in which the domain wall density is controlled by a fugacity and show that it maps on to the truncated inverse distance squared Ising model. The generalised model shows a mixed order phase transition, with the regular CD model (which exhibits FDPO) corresponding to the critical point.
37
López-Couso, María José, e Belén Méndez-Naya. "From clause to pragmatic marker". Journal of Historical Pragmatics 15, n. 1 (28 febbraio 2014): 36–61. http://dx.doi.org/10.1075/jhp.15.1.03lop.
Abstract (sommario):
Propositional attitude predicates, such as English think, guess and seem, occur parenthetically in many languages. In this article we pay attention to a group of propositional attitude predicates which can be labelled epistemic/evidential, namely appear, look, seem and sound, and which, in addition to degree of certainty, also give an indication of the evidential source. In this study we describe the different parentheticals available with these verbs, paying special attention to like-parentheticals (e.g. Going to be a big one, looks like), a development characteristic of American English. Using data from the Corpus of Historical American English (COHA, Davies 2010-) and the Corpus of Contemporary American English (COCA, Davies 2008-), we explore whether these developments can be conceptualised in terms of grammaticalisation and (inter)subjectification. In the structural domain, these parentheticals show fixation, decategorialisation and fusion. In the semantic-pragmatic domain, they show signs of generalisation of meaning and increased (inter)subjectivity.
38
Steinberg, Benjamin. "Inverse semigroup homomorphisms via partial group actions". Bulletin of the Australian Mathematical Society 64, n. 1 (agosto 2001): 157–68. http://dx.doi.org/10.1017/s0004972700019778.
Abstract (sommario):
This papar constructs all homomorphisms of inverse semigroups which factor through an E-unitary inverse semigroup; the construction is in terms of a semilattice component and a group component. It is shown that such homomorphisms have a unique factorisation βα with α preserving the maximal group image, β idempotent separating, and the domain I of β E-unitary; moreover, the P-representation of I is explicitly constructed. This theory, in particular, applies whenever the domain or codomain of a homomorphism is E-unitary. Stronger results are obtained for the case of F-inverse monoids.Special cases of our results include the P-theorem and the factorisation theorem for homomorphisms from E-unitary inverse semigroups (via idempotent pure followed by idempotent separating). We also deduce a criterion of McAlister–Reilly for the existence of E-unitary covers over a group, as well as a generalisation to F-inverse covers, allowing a quick proof that every inverse monoid has an F-inverse cover.
39
Fokas, Athanasios S., e Konstantinos Kalimeris. "Water waves with moving boundaries". Journal of Fluid Mechanics 832 (26 ottobre 2017): 641–65. http://dx.doi.org/10.1017/jfm.2017.681.
Abstract (sommario):
The unified transform, also known as the Fokas method, provides a powerful methodology for studying boundary value problems. Employing this methodology, we analyse inviscid, irrotational, two-dimensional water waves in a bounded domain, and in particular we study the generation of waves by a moving piecewise horizontal bottom, as it occurs in tsunamis. We show that this problem is characterised by two equations which involve only first-order derivatives. It is argued that under the assumptions of ‘small amplitude waves’ but not of ‘long waves’, the above two equations can be treated numerically via a recently introduced numerical technique for elliptic partial differential equations in a polygonal domain. In the particular case that the moving bottom is horizontal and under the assumption of ‘small amplitude waves’, but not of ‘long waves’, these equations yield a non-local generalisation of the Boussinesq system. Furthermore, under the additional assumption of ‘long waves’ the above system yields a Boussinesq-type system, which however includes the effect of the moving boundary.
40
Alzubaidi, Laith, Asma Salhi, Mohammed A.Fadhel, Jinshuai Bai, Freek Hollman, Kristine Italia, Roberto Pareyon et al. "Trustworthy deep learning framework for the detection of abnormalities in X-ray shoulder images". PLOS ONE 19, n. 3 (11 marzo 2024): e0299545. http://dx.doi.org/10.1371/journal.pone.0299545.
Abstract (sommario):
Musculoskeletal conditions affect an estimated 1.7 billion people worldwide, causing intense pain and disability. These conditions lead to 30 million emergency room visits yearly, and the numbers are only increasing. However, diagnosing musculoskeletal issues can be challenging, especially in emergencies where quick decisions are necessary. Deep learning (DL) has shown promise in various medical applications. However, previous methods had poor performance and a lack of transparency in detecting shoulder abnormalities on X-ray images due to a lack of training data and better representation of features. This often resulted in overfitting, poor generalisation, and potential bias in decision-making. To address these issues, a new trustworthy DL framework has been proposed to detect shoulder abnormalities (such as fractures, deformities, and arthritis) using X-ray images. The framework consists of two parts: same-domain transfer learning (TL) to mitigate imageNet mismatch and feature fusion to reduce error rates and improve trust in the final result. Same-domain TL involves training pre-trained models on a large number of labelled X-ray images from various body parts and fine-tuning them on the target dataset of shoulder X-ray images. Feature fusion combines the extracted features with seven DL models to train several ML classifiers. The proposed framework achieved an excellent accuracy rate of 99.2%, F1Score of 99.2%, and Cohen’s kappa of 98.5%. Furthermore, the accuracy of the results was validated using three visualisation tools, including gradient-based class activation heat map (Grad CAM), activation visualisation, and locally interpretable model-independent explanations (LIME). The proposed framework outperformed previous DL methods and three orthopaedic surgeons invited to classify the test set, who obtained an average accuracy of 79.1%. The proposed framework has proven effective and robust, improving generalisation and increasing trust in the final results.
41
Ibarra, Emiro J., Julián D. Arias-Londoño, Matías Zañartu e Juan I. Godino-Llorente. "Towards a Corpus (and Language)-Independent Screening of Parkinson’s Disease from Voice and Speech through Domain Adaptation". Bioengineering 10, n. 11 (15 novembre 2023): 1316. http://dx.doi.org/10.3390/bioengineering10111316.
Abstract (sommario):
End-to-end deep learning models have shown promising results for the automatic screening of Parkinson’s disease by voice and speech. However, these models often suffer degradation in their performance when applied to scenarios involving multiple corpora. In addition, they also show corpus-dependent clusterings. These facts indicate a lack of generalisation or the presence of certain shortcuts in the decision, and also suggest the need for developing new corpus-independent models. In this respect, this work explores the use of domain adversarial training as a viable strategy to develop models that retain their discriminative capacity to detect Parkinson’s disease across diverse datasets. The paper presents three deep learning architectures and their domain adversarial counterparts. The models were evaluated with sustained vowels and diadochokinetic recordings extracted from four corpora with different demographics, dialects or languages, and recording conditions. The results showed that the space distribution of the embedding features extracted by the domain adversarial networks exhibits a higher intra-class cohesion. This behaviour is supported by a decrease in the variability and inter-domain divergence computed within each class. The findings suggest that domain adversarial networks are able to learn the common characteristics present in Parkinsonian voice and speech, which are supposed to be corpus, and consequently, language independent. Overall, this effort provides evidence that domain adaptation techniques refine the existing end-to-end deep learning approaches for Parkinson’s disease detection from voice and speech, achieving more generalizable models.
42
Nti, Isaac Kofi, Owusu Nyarko-Boateng, Felix Adebayo Adekoya e Benjamin Asubam Weyori. "An empirical assessment of different kernel functions on the performance of support vector machines". Bulletin of Electrical Engineering and Informatics 10, n. 6 (1 dicembre 2021): 3403–11. http://dx.doi.org/10.11591/eei.v10i6.3046.
Abstract (sommario):
Artificial intelligence (AI) and machine learning (ML) have influenced every part of our day-to-day activities in this era of technological advancement, making a living more comfortable on the earth. Among the several AI and ML algorithms, the support vector machine (SVM) has become one of the most generally used algorithms for data mining, prediction and other (AI and ML) activities in several domains. The SVM’s performance is significantly centred on the kernel function (KF); nonetheless, there is no universal accepted ground for selecting an optimal KF for a specific domain. In this paper, we investigate empirically different KFs on the SVM performance in various fields. We illustrated the performance of the SVM based on different KF through extensive experimental results. Our empirical results show that no single KF is always suitable for achieving high accuracy and generalisation in all domains. However, the gaussian radial basis function (RBF) kernel is often the default choice. Also, if the KF parameters of the RBF and exponential RBF are optimised, they outperform the linear and sigmoid KF based SVM method in terms of accuracy. Besides, the linear KF is more suitable for the linearly separable dataset.
43
BÉRCZES, ATTILA. "Effective results for unit points on curves over finitely generated domains". Mathematical Proceedings of the Cambridge Philosophical Society 158, n. 2 (13 gennaio 2015): 331–53. http://dx.doi.org/10.1017/s0305004114000693.
Abstract (sommario):
AbstractLet A be a commutative domain of characteristic 0 which is finitely generated over ℤ as a ℤ-algebra. Denote by A* the unit group of A and by K the algebraic closure of the quotient field K of A. We shall prove effective finiteness results for the elements of the set \begin{equation*} \mathcal{C}:=\{ (x,y)\in (A^*)^2 | F(x,y)=0 \} \end{equation*} where F(X, Y) is a non-constant polynomial with coefficients in A which is not divisible over K by any polynomial of the form XmYn - α or Xm - α Yn, with m, n ∈ ℤ⩾0, max(m, n) > 0, α ∈ K*. This result is a common generalisation of effective results of Evertse and Győry [12] on S-unit equations over finitely generated domains, of Bombieri and Gubler [5] on the equation F(x, y) = 0 over S-units of number fields, and it is an effective version of Lang's general but ineffective theorem [20] on this equation over finitely generated domains. The conditions that A is finitely generated and F is not divisible by any polynomial of the above type are essentially necessary.
44
Essegbey, James. "The “basic locative construction” in Gbe languages and Surinamese creoles". Journal of Pidgin and Creole Languages 20, n. 2 (29 novembre 2005): 229–67. http://dx.doi.org/10.1075/jpcl.20.2.02ess.
Abstract (sommario):
This paper compares the conceptualisation and expression of topological relations in Surinamese creoles with that of Gbe languages (which were part of the substrate) and English (the superstrate). It investigates the components of the Basic Locative Construction (BLC), i.e. the most neutral construction that is used to code topology, and the type of situations for which the BLC is used in the languages. It shows that the BLC in the creole and Gbe languages has a locative phrase which is made up of a noun phrase that expresses the Ground and a spatial element that expresses the Search Domain i.e. the specific part of the Ground where the Figure is located. The locative phrase in the creoles also has a preposition but this does not contribute to its spatial meaning. By contrast, English has a locative phrase which is made up of a preposition that expresses the Relation between the Figure and the Ground, and the Search-Domain information. The paper concludes that the Suriname creoles display a strong substrate influence in this spatial domain. There are some differences, which can be attributed to gradual Dutch influence and generalisation on the part of the creoles.
45
Carozzi, T. D., A. M. Buckley e M. P. Gough. "Instantaneous local wave vector estimation from multi-spacecraft measurements using few spatial points". Annales Geophysicae 22, n. 7 (14 luglio 2004): 2633–41. http://dx.doi.org/10.5194/angeo-22-2633-2004.
Abstract (sommario):
Abstract. We introduce a technique to determine instantaneous local properties of waves based on discrete-time sampled, real-valued measurements from 4 or more spatial points. The technique is a generalisation to the spatial domain of the notion of instantaneous frequency used in signal processing. The quantities derived by our technique are closely related to those used in geometrical optics, namely the local wave vector and instantaneous phase velocity. Thus, this experimental technique complements ray-tracing. We provide example applications of the technique to electric field and potential data from the EFW instrument on Cluster. Cluster is the first space mission for which direct determination of the full 3-dimensional local wave vector is possible, as described here.
46
Gobet, Fernand. "The 35th Cognitive Psychology Section Annual Conference". Cognitive Psychology Bulletin 1, n. 4 (2019): 4–9. http://dx.doi.org/10.53841/bpscog.2019.1.4.4.
Abstract (sommario):
Chess players, musicians and video-game players tend to be more intelligent than individuals not engaged in these activities. It has thus been proposed that practising these activities improves cognitive ability and fosters educational achievements. Similar claims have been made with respect to other activities such as working memory training and brain training. This talk will review several meta-analyses performed to address these claims. It will also discuss the results of a second-order meta-analysis (a meta-meta-analysis) that examines all these domains simultaneously. The results are very consistent across domains: whilst there is evidence for near transfer (i.e. transfer to similar tasks), there is very little evidence for far transfer (i.e. transfer of a set of skills between domains only weakly related to each other). When placebo effects and publication biases are controlled for, the overall effect size for far transfer is essentially zero. These results cast serious doubts on the effectiveness of cognitive training for improving overall cognitive ability. They are also theoretically important: they support theories of learning and expertise such as chunking and template theories that emphasise domain-specific knowledge, and strongly suggest that the lack of generalisation of skills acquired by training is an invariant of human cognition.
47
Xie, Y., K. Schindler, J. Tian e X. X. Zhu. "EXPLORING CROSS-CITY SEMANTIC SEGMENTATION OF ALS POINT CLOUDS". International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences XLIII-B2-2021 (28 giugno 2021): 247–54. http://dx.doi.org/10.5194/isprs-archives-xliii-b2-2021-247-2021.
Abstract (sommario):
Abstract. Deep learning models achieve excellent semantic segmentation results for airborne laser scanning (ALS) point clouds, if sufficient training data are provided. Increasing amounts of annotated data are becoming publicly available thanks to contributors from all over the world. However, models trained on a specific dataset typically exhibit poor performance on other datasets. I.e., there are significant domain shifts, as data captured in different environments or by distinct sensors have different distributions. In this work, we study this domain shift and potential strategies to mitigate it, using two popular ALS datasets: the ISPRS Vaihingen benchmark from Germany and the LASDU benchmark from China. We compare different training strategies for cross-city ALS point cloud semantic segmentation. In our experiments, we analyse three factors that may lead to domain shift and affect the learning: point cloud density, LiDAR intensity, and the role of data augmentation. Moreover, we evaluate a well-known standard method of domain adaptation, deep CORAL (Sun and Saenko, 2016). In our experiments, adapting the point cloud density and appropriate data augmentation both help to reduce the domain gap and improve segmentation accuracy. On the contrary, intensity features can bring an improvement within a dataset, but deteriorate the generalisation across datasets. Deep CORAL does not further improve the accuracy over the simple adaptation of density and data augmentation, although it can mitigate the impact of improperly chosen point density, intensity features, and further dataset biases like lack of diversity.
48
Wu, Guile, e Shaogang Gong. "Decentralised Learning from Independent Multi-Domain Labels for Person Re-Identification". Proceedings of the AAAI Conference on Artificial Intelligence 35, n. 4 (18 maggio 2021): 2898–906. http://dx.doi.org/10.1609/aaai.v35i4.16396.
Abstract (sommario):
Deep learning has been successful for many computer vision tasks due to the availability of shared and centralised large-scale training data. However, increasing awareness of privacy concerns poses new challenges to deep learning, especially for human subject related recognition such as person re-identification (Re-ID). In this work, we solve the Re-ID problem by decentralised learning from non-shared private training data distributed at multiple user sites of independent multi-domain label spaces. We propose a novel paradigm called Federated Person Re-Identification (FedReID) to construct a generalisable global model (a central server) by simultaneously learning with multiple privacy-preserved local models (local clients). Specifically, each local client receives global model updates from the server and trains a local model using its local data independent from all the other clients. Then, the central server aggregates transferrable local model updates to construct a generalisable global feature embedding model without accessing local data so to preserve local privacy. This client-server collaborative learning process is iteratively performed under privacy control, enabling FedReID to realise decentralised learning without sharing distributed data nor collecting any centralised data. Extensive experiments on ten Re-ID benchmarks show that FedReID achieves compelling generalisation performance beyond any locally trained models without using shared training data, whilst inherently protects the privacy of each local client. This is uniquely advantageous over contemporary Re-ID methods.
49
Saranti, Anna, Miroslav Hudec, Erika Mináriková, Zdenko Takác̆, Udo Großschedl, Christoph Koch, Bastian Pfeifer, Alessa Angerschmid e Andreas Holzinger. "Actionable Explainable AI (AxAI): A Practical Example with Aggregation Functions for Adaptive Classification and Textual Explanations for Interpretable Machine Learning". Machine Learning and Knowledge Extraction 4, n. 4 (27 ottobre 2022): 924–53. http://dx.doi.org/10.3390/make4040047.
Abstract (sommario):
In many domains of our daily life (e.g., agriculture, forestry, health, etc.), both laymen and experts need to classify entities into two binary classes (yes/no, good/bad, sufficient/insufficient, benign/malign, etc.). For many entities, this decision is difficult and we need another class called “maybe”, which contains a corresponding quantifiable tendency toward one of these two opposites. Human domain experts are often able to mark any entity, place it in a different class and adjust the position of the slope in the class. Moreover, they can often explain the classification space linguistically—depending on their individual domain experience and previous knowledge. We consider this human-in-the-loop extremely important and call our approach actionable explainable AI. Consequently, the parameters of the functions are adapted to these requirements and the solution is explained to the domain experts accordingly. Specifically, this paper contains three novelties going beyond the state-of-the-art: (1) A novel method for detecting the appropriate parameter range for the averaging function to treat the slope in the “maybe” class, along with a proposal for a better generalisation than the existing solution. (2) the insight that for a given problem, the family of t-norms and t-conorms covering the whole range of nilpotency is suitable because we need a clear “no” or “yes” not only for the borderline cases. Consequently, we adopted the Schweizer–Sklar family of t-norms or t-conorms in ordinal sums. (3) A new fuzzy quasi-dissimilarity function for classification into three classes: Main difference, irrelevant difference and partial difference. We conducted all of our experiments with real-world datasets.
50
Stoltenberg-Hansen, V., e J. V. Tucker. "Complete local rings as domains". Journal of Symbolic Logic 53, n. 2 (giugno 1988): 603–24. http://dx.doi.org/10.1017/s0022481200028498.
Abstract (sommario):
Contents: Introduction. §1: Computable rings and modules. §2: Ideal membership relation. §3: Effective structured domains. §4: Completion of a local ring as a domain. §5: The recursive completion. Epilogue. References.Introduction. Completion is an important general mathematical device. Often, but not always, a completion takes the following form. Let A be a topological algebraic structure whose topology is derived from a metric. For A, a topological algebra  and an embedding i: A →  are constructed such that  is a complete metric space in which A is densely embedded by i. The long list of structures for which such completions exist begins with Cantor's construction of the real number field and includes objects like the p-adic integers, Baire space, and Boolean algebras. In Bourbaki [6] a careful and thorough account of completions for arbitrary topological groups and fields is given, for which it is important to note that the topological structures need not be metrizable, but must possess a uniformity.The effectiveness of the completion process of a computable structure A cannot be readily studied using the tools of computable algebra, simply because the resulting structure  is almost invariably uncountable. However, in particular cases, it has been possible to define and study the substructure Ak of computable elements of Â; this has been done for the structures mentioned above, starting with the field of recursive real numbers.In this paper we analyse the effectivity of the completion of a local ring R. We do this using structured Scott-Ershov domains. Our study may be considered as a prototype containing methods applicable to a broad class of completions, including all the examples mentioned above, except for the real number field, which needs a generalisation of the domain concept.A Scott-Ershov domain D formalises how a set Dt of possibly “infinite” elements, called total elements, is constructed from a set Dc of “finite” elements, called compact elements. This is achieved by means of an approximation ordering which determines a topology on D and, in particular, on Dt. Our methodology is to associate to a given topological algebra A a structured domain D(A) such that the total elements D(A)t form a topological algebra topologically isomorphic to A. In such circumstances A is said to be domain definable by D(A). The theory of computability for domains is now applied to study the effectivity of the topological algebra A.