Vertical Federated Learning (VFL) is becoming a standard collaborative learning paradigm with various practical applications. Randomness is essential to enhancing privacy in VFL, but introducing too much external rand...
ISBN:
(纸本)9783031708893;9783031708909
Vertical Federated Learning (VFL) is becoming a standard collaborative learning paradigm with various practical applications. Randomness is essential to enhancing privacy in VFL, but introducing too much external randomness often leads to an intolerable performance loss. Instead, as it was demonstrated for other federated learning settings, leveraging internal randomness - as provided by variational autoencoders (VAEs) -can be beneficial. However, the resulting privacy has never been quantified so far, nor has the approach been investigated for VFL. We therefore propose a novel differential privacy (DP) estimate, denoted as distance-based empirical local differential privacy (dELDP). It allows us to empirically bound DP parameters of models or model components, quantifying the internal randomness with appropriate distance and sensitivity metrics. We apply dELDP to investigate the DP of VAEs and observe values up to epsilon approximate to 6.4 and delta = 2(-32). Based on this, to link the dELDP parameters to the privacy of VAE-including VFL systems in practice, we conduct comprehensive experiments on the robustness against state-of-the-art privacy attacks. The results illustrate that the VAE system is robust against feature reconstruction attacks and outperforms other privacy-enhancing methods for VFL, especially when the adversary holds 75% of the features during label inference attacks.
In the contemporary digital landscape, the pervasive and far-reaching impact of online social networks is indisputable. Prominent platforms such as Instagram, Facebook, and Twitter frequently grapple with the persiste...
ISBN:
(数字)9783031599330
ISBN:
(纸本)9783031599323;9783031599330
In the contemporary digital landscape, the pervasive and far-reaching impact of online social networks is indisputable. Prominent platforms such as Instagram, Facebook, and Twitter frequently grapple with the persistent challenge of distinguishing between registered profiles and genuinely engaged users, resulting in a noticeable surge in the prevalence of counterfeit or dormant accounts. This situation underscores the compelling necessity to accurately differentiate between authentic and misleading user profiles. The primary objective of this investigation is to introduce an innovative approach to profile validation. This unique method astutely leverages state-of-the-art bio-inspired algorithms while circumventing traditional machine learning techniques. The empirical results are notably convincing, consistently achieving a high level of accuracy in classification tests conducted on the provided datasets.
We are entering a new era in which software systems are becoming more and more complex and larger. So, the composition of such systems is becoming infeasible by manual means. To address this challenge, self-organising...
ISBN:
(纸本)9783031646256;9783031646263
We are entering a new era in which software systems are becoming more and more complex and larger. So, the composition of such systems is becoming infeasible by manual means. To address this challenge, self-organising software models represent a promising direction since they allow the (bottom-up) emergence of complex computational structures from simple rules. In this paper, we propose an abstract machine, called the composition machine, which allows the definition and the execution of such models. Unlike typical abstract machines, our proposal does not compute individual programs but enables the emergence of multiple programs at once. We particularly present the machine's semantics and demonstrate its operation with well-known rules from the realm of Boolean logic and elementary cellular automata.
Pseudo-relevance feedback (PRF) can enhance average retrieval effectiveness over a sufficiently large number of queries. However, PRF often introduces a drift into the original information need, thus hurting the retri...
ISBN:
(纸本)9783031560590;9783031560606
Pseudo-relevance feedback (PRF) can enhance average retrieval effectiveness over a sufficiently large number of queries. However, PRF often introduces a drift into the original information need, thus hurting the retrieval effectiveness of several queries. While a selective application of PRF can potentially alleviate this issue, previous approaches have largely relied on unsupervised or feature-based learning to determine whether a query should be expanded. In contrast, we revisit the problem of selective PRF from a deep learning perspective, presenting a model that is entirely data-driven and trained in an end-to-end manner. The proposed model leverages a transformer-based bi-encoder architecture. Additionally, to further improve retrieval effectiveness with this selective PRF approach, we make use of the model's confidence estimates to combine the information from the original and expanded queries. In our experiments, we apply this selective feedback on a number of different combinations of ranking and feedback models, and show that our proposed approach consistently improves retrieval effectiveness for both sparse and dense ranking models, with the feedback models being either sparse, dense or generative.
In model-driven engineering, runtime monitoring of systems with complex dynamic structures is typically performed via a runtime model capturing a snapshot of the system state: the model is represented as a graph and p...
ISBN:
(纸本)9783031572586;9783031572593
In model-driven engineering, runtime monitoring of systems with complex dynamic structures is typically performed via a runtime model capturing a snapshot of the system state: the model is represented as a graph and properties of interest as graph queries which are evaluated over the model online. For temporal properties, history-aware runtime models encode a trace of timestamped snapshots, which is monitored via temporal graph queries. In this case, the query evaluation needs to consider that a trace may be incomplete, thus future changes to the model may affect current answers. So far there is no formal foundation for query-based monitoring over runtime models encoding incomplete traces. In this paper, we present a systematic and formal treatment of incomplete traces. First, we introduce a new definite semantics for a first-order temporal graph logic which only returns answers if no future change to the model will affect them. Then, we adjust the query evaluation semantics of a querying approach we previously presented, which is based on this logic, to the definite semantics of the logic. Lastly, we enable the approach to keep to its efficient query evaluation technique, while returning (the more costly) definite answers.
We study the natural extended-variable formulation for the disjunction of n+1 polytopes in R-d. We demonstrate that the convex hull D in the natural extended-variable space Rd+n is given by full optimal big-M lifting ...
ISBN:
(纸本)9783031609237;9783031609244
We study the natural extended-variable formulation for the disjunction of n+1 polytopes in R-d. We demonstrate that the convex hull D in the natural extended-variable space Rd+n is given by full optimal big-M lifting (i) when d <= 2 (and that it is not generally true for d >= 3), and also (ii) when the polytopes are all axis-aligned hyper-rectangles. We give further results on the polyhedral structure of D, emphasizing the role of full optimal big-M lifting.
This paper provides an overview of spherical blossoms, called splossoms, and some of its implications. The blossom of a polynomial is a multi-affine function of euclidean space with the same number of variables as the...
ISBN:
(纸本)9783031500787;9783031500770
This paper provides an overview of spherical blossoms, called splossoms, and some of its implications. The blossom of a polynomial is a multi-affine function of euclidean space with the same number of variables as the degree of the polynomial. It provides many insights to the polynomial and simplifies methods not otherwise apparent. One example is the de Casteljau algorithm for computing and subdividing a Bezier curve. This report describes a blossom for a parametric de Casteljau-like curve on the sphere, leading to similar insights and simplification of algorithms on the sphere. Two earlier such methods are the well-known SLERP and SQUAD interpolations of points on the sphere. These methods are re-formulated with our new concept, the splossom, which plays the role of a blossom in spherical space. Some of its implications are briefly sketched to illustrate its potential. The splossom itself is neatly described in terms of spinors in Geometric Algebra. This development follows the Geometric Algebra approach and points to considerable further research within its broad vista.
The ability to measure gene expression at single-cell resolution has elevated our understanding of how biological features emerge from complex and interdependent networks at molecular, cellular, and tissue scales. As ...
ISBN:
(纸本)9781071639887;9781071639894
The ability to measure gene expression at single-cell resolution has elevated our understanding of how biological features emerge from complex and interdependent networks at molecular, cellular, and tissue scales. As technologies have evolved that complement scRNAseq measurements with things like single-cell proteomic, epigenomic, and genomic information, it becomes increasingly apparent how much biology exists as a product of multimodal regulation. Biological processes such as transcription, translation, and post-translational or epigenetic modification impose both energetic and specific molecular demands on a cell and are therefore implicitly constrained by the metabolic state of the cell. While metabolomics is crucial for defining a holistic model of any biological process, the chemical heterogeneity of the metabolome makes it particularly difficult to measure, and technologies capable of doing this at single-cell resolution are far behind other multiomics modalities. To address these challenges, we present GEFMAP (Gene Expression-based Flux Mapping and Metabolic Pathway Prediction), a method based on geometric deep learning for predicting flux through reactions in a global metabolic network using transcriptomics data, which we ultimately apply to scRNAseq. GEFMAP leverages the natural graph structure of metabolic networks to learn both a biological objective for each cell and estimate a mass-balanced relative flux rate for each reaction in each cell using novel deep learning models.
Nowadays Discrete Event Systems (DESs) require complex and large models, for which distributed simulation engines become, in practice, the tools used to understand and analyze their behavior. The feasibility and effic...
ISBN:
(数字)9783031614330
ISBN:
(纸本)9783031614323;9783031614330
Nowadays Discrete Event Systems (DESs) require complex and large models, for which distributed simulation engines become, in practice, the tools used to understand and analyze their behavior. The feasibility and efficiency of a distributed simulation of these large-scale models is strongly dependent of the information that can be obtained from the models, previously to the simulation process itself. This information can give assistance to the generation of an initial partition of the model, allowing a well balanced workload among the individual simulation engines deployed, or in the generation of the predicates to be evaluated in order to determine the enabling of transitions;or the computation of look-ahead information in conservative strategies of distributed simulation. Petri nets allow to obtain information from the structure that can be used to advance conclusions or properties about the course of a simulation. This information can be usefull either independently of the considered initial marking, or parameterised by its initial choice. This structural information can be obtained in modelling phase, completed in simulation time and re-elaborated from the simulation results, and therefore associated to the model or modules of the model in such a way that can be harnessed in further simulations where these nets will be used. Last but no least, the maintenance of the structure of the Petri net during the simulation (in an interpreted simulation instead of a compiled one) allows to make load balancing during the simulation or to federate with legacy simulators, in an easier way than using other kind of specification models or simulation schemes.
Named entity recognition (NER) is an important component of many information extraction and linking pipelines. The task is especially challenging in a low-resource scenario, where there is very limited amount of high ...
ISBN:
(数字)9783031539695
ISBN:
(纸本)9783031539688;9783031539695
Named entity recognition (NER) is an important component of many information extraction and linking pipelines. The task is especially challenging in a low-resource scenario, where there is very limited amount of high quality annotated data. In this paper we benchmark machine learning approaches for NER that may be very effective in such cases, and compare their performance in a novel application;information extraction of research infrastructure from scientific manuscripts. We explore approaches such as incorporating Contrastive Learning (CL), as well as Conditional Random Fields (CRF) weights in BERT-based architectures and demonstrate experimentally that such combinations are very efficient in few-shot learning set-ups, verifying similar findings that have been reported in other areas of NLP, as well as computer Vision. More specifically, we show that the usage of CRF weights in BERT-based architectures achieves noteworthy improvements in the overall NER task by approximately 12%, and that in few-shot setups the effectiveness of CRF weights is much higher in smaller training sets.
暂无评论