Search

Traditional machine translation methods typically involve training models directly on large parallel corpora, with limited emphasis on specialized terminology. However, In specialized fields such as patent, finance, or biomedical domains, terminology is crucial for translation, with many terms that needs to be translated following agreed-upon conventions. In this paper we introduce a methodology that efficiently trains models with a smaller amount of data while preserving the accuracy of terminology translation. We achieve this through a systematic process of term extraction and glossary creation using the Trie Tree algorithm, followed by data reconstruction to teach the LLM how to integrate these specialized terms. This methodology enhances the model's ability to handle specialized terminology and ensures high-quality translations, particularly in fields where term consistency is crucial. Our approach has demonstrated exceptional performance, achieving the highest translation score among participants in the WMT patent task to date, showcasing its effectiveness and broad applicability in specialized translation domains where general methods often fall short., Comment: Accepted to WMT 2024

The rapid advance of quantum computing has highlighted the need for scalable and efficient software infrastructures to fully exploit its potential. While current quantum processors have significant scalability problems due to the limited number of qubits on each chip, distributed quantum computing offers a promising solution by networking multiple Quantum Processing Units (QPUs). To support this paradigm, robust Intermediate Representations (IRs) are crucial for translating high-level quantum algorithms into executable instructions across distributed systems. This paper introduces NetQIR, an extension of Microsoft's Quantum Intermediate Representation (QIR), specifically designed for distributed quantum computing. NetQIR is designed to meet the specific needs of distributed quantum systems by incorporating functions to manage quantum and classical communications between QPUs. The main objective is to facilitate the development of new distributed compilers by improving the integration and execution of quantum programmes in a distributed infrastructure, taking advantage of modular architectures to improve scalability. By extending QIR to support distributed quantum computing, NetQIR aims to complement and add capabilities to an already supported quantum IR and, at the same time, take advantage of the tools previously created for QIR. Throughout this paper the specification of the intermediate representation is introduced, including the basic instructions necessary to enable distributed quantum computing in an abstract form independent of the target machine.

Curry-Howard correspondences between Linear Logic (LL) and session types provide a firm foundation for concurrent processes. As the correspondences hold for intuitionistic and classic versions of LL (ILL and CLL), we obtain two different families of type systems for concurrency. An open question remains: how do these two families exactly relate to each other? Based upon a translation from CLL to ILL due to Laurent (2018), we provide two complementary answers, in the form of full abstraction results based on a typed observational equivalence due to Atkey (2017). Our results elucidate hitherto missing formal links between seemingly related yet different type systems for concurrency., Comment: Full version, 19 pages + appendices

In order to enhance safety, nuclear reactors in the design phase consider natural circulation as a mean to remove residual power. The simulation of this passive mechanism must be qualified between the validation range and the scope of utilization (reactor case), introducing potential physical and numerical distortion effects. In this study, we simulate the flow of liquid sodium using the TrioCFD code, employing both higher-fidelity (HF) LES and lower-fidelity (LF) URANS models. We tackle respectively numerical uncertainties through the Grid Convergence Index method, and physical modelling uncertainties through the Polynomial Chaos Expansion method available on the URANIE platform. HF simulations are shown to exhibit a strong resilience to physical distortion effects, with numerical uncertainties being intricately correlated. Conversely, the LF approach, the only one applicable at the reactor scale, is likely to present a reduced predictability. If so, the HF approach should be effective in pinpointing the LF weaknesses: the concept of scaling uncertainty is inline introduced as the growth of the LF simulation uncertainty associated with distortion effects. Thus, the paper outlines that a specific methodology within the BEPU framework - leveraging both HF and LF approaches - could pragmatically enable correlating distortion effects with scaling uncertainty, thereby providing a metric principle.

The emerging field of quantum computing has shown it might change how we process information by using the unique principles of quantum mechanics. As researchers continue to push the boundaries of quantum technologies to unprecedented levels, distributed quantum computing raises as an obvious path to explore with the aim of boosting the computational power of current quantum systems. This paper presents a comprehensive survey of the current state of the art in the distributed quantum computing field, exploring its foundational principles, landscape of achievements, challenges, and promising directions for further research. From quantum communication protocols to entanglement-based distributed algorithms, each aspect contributes to the mosaic of distributed quantum computing, making it an attractive approach to address the limitations of classical computing. Our objective is to provide an exhaustive overview for experienced researchers and field newcomers.

Session types are a typed approach to message-passing concurrency, where types describe sequences of intended exchanges over channels. Session type systems have been given strong logical foundations via Curry-Howard correspondences with linear logic, a resource-aware logic that naturally captures structured interactions. These logical foundations provide an elegant framework to specify and (statically) verify message-passing processes. In this paper, we rigorously compare different type systems for concurrency derived from the Curry-Howard correspondence between linear logic and session types. We address the main divide between these type systems: the classical and intuitionistic presentations of linear logic. Over the years, these presentations have given rise to separate research strands on logical foundations for concurrency; the differences between their derived type systems have only been addressed informally. To formally assess these differences, we develop $\pi\mathsf{ULL}$, a session type system that encompasses type systems derived from classical and intuitionistic interpretations of linear logic. Based on a fragment of Girard's Logic of Unity, $\pi\mathsf{ULL}$ provides a basic reference framework: we compare existing session type systems by characterizing fragments of $\pi\mathsf{ULL}$ that coincide with classical and intuitionistic formulations. We analyze the significance of our characterizations by considering the locality principle (enforced by intuitionistic interpretations but not by classical ones) and forms of process composition induced by the interpretations., Comment: Preprint to appear in JLAMP; revised/extended version of https://doi.org/10.4204/EPTCS.314.1

Tele-operated Driving (ToD) is a challenging use case for mobile network operators. Video captured by the built-in vehicle cameras must be streamed meeting a latency requirement of 5 ms with a 99.999% reliability. Although 5G offers high bandwidth, ultra-low latencies and high reliability; ToD service requirements are violated due to bad channel conditions. Ignoring the channel state may lead to over-estimate the number of ToD vehicles that can meet the service requirements, hence comprising the vehicle security. To fill this gap, in this letter we propose TOVAC, an algorithm that guarantees ToD service requirements by taking adequate admission control and routing decisions. This is achieved by using a channel-based capacity graph that determines the maximum number of vehicles that can be tele-operated in any road section. We evaluate TOVAC considering cellular deployments from Turin and show that, unlike a state of the art solution, TOVAC guarantees the ToD service requirements., Comment: 5 pages, 5 figures

Road safety is the main motivation for Cooperative Intelligent Transport Systems (C-ITS) in general, and vehicular communications (V2X) technology in particular. The V2X-based Vulnerable Road User (VRU) protection is an approach that relies on the persistent broadcasting of "beacon" awareness messages by a VRU mobile device. To this end the European Telecommunications Standards Institute (ETSI) has specified the Vulnerable Road User Awareness Message (VAM) as well as the overall ITS-G5 protocol stack enabling a variety of the V2X applications. This article studies how often pedestrians (a type of VRU) should check their position to issue a VAM. To that end, we characterize the rate at which pedestrians generate VAMs leveraging a recognized mobility model, and formulate an optimization problem to minimize the time elapsed between VAMs. We propose an algorithm to solve the problem in 802.11p and assess its accuracy through numerical and simulation campaigns. Results evidence the accuracy of our VAM rate characterization, and evidence that we decrease ETSI positioning sampling rate by more than 30%. On top, our solution decreases the time between VAMs, and increases the packet delivery ratio. In other words, our approach increases the pedestrians safety while reducing the battery consumption of mobile devices., Comment: 17 pages, 11 figures

The Spanish language is one of the top 5 spoken languages in the world. Nevertheless, finding resources to train or evaluate Spanish language models is not an easy task. In this paper we help bridge this gap by presenting a BERT-based language model pre-trained exclusively on Spanish data. As a second contribution, we also compiled several tasks specifically for the Spanish language in a single repository much in the spirit of the GLUE benchmark. By fine-tuning our pre-trained Spanish model, we obtain better results compared to other BERT-based models pre-trained on multilingual corpora for most of the tasks, even achieving a new state-of-the-art on some of them. We have publicly released our model, the pre-training data, and the compilation of the Spanish benchmarks., Comment: Published as workshop paper at Practical ML for Developing Countries Workshop @ ICLR 2020

Termination is a central property in sequential programming models: a term is terminating if all its reduction sequences are finite. Termination is also important in concurrency in general, and for message-passing programs in particular. A variety of type systems that enforce termination by typing have been developed. In this paper, we rigorously compare several type systems for $\pi$-calculus processes from the unifying perspective of termination. Adopting session types as reference framework, we consider two different type systems: one follows Deng and Sangiorgi's weight-based approach; the other is Caires and Pfenning's Curry-Howard correspondence between linear logic and session types. Our technical results precisely connect these very different type systems, and shed light on the classes of client/server interactions they admit as correct.

Industry 4.0 applications impose the challenging demand of delivering packets with bounded latencies via a wireless network. This is further complicated if the network is not dedicated to the time critical application. In this paper we use network calculus analysis to derive closed form expressions of latency bounds for time critical traffic when 802.11 Target Wake Time (TWT) and 802.1Qbv work together in a shared 802.11 network., Comment: 3 pages, 3 figures, workshop submission

We present a framework for the distributed monitoring of networks of components that coordinate by message-passing, following multiparty session protocols specified as global types. We improve over prior works by (i) supporting components whose exact specification is unknown ("blackboxes") and (ii) covering protocols that cannot be analyzed by existing techniques. We first give a procedure for synthesizing monitors for blackboxes from global types, and precisely define when a blackbox correctly satisfies its global type. Then, we prove that monitored blackboxes are sound (they correctly follow the protocol) and transparent (blackboxes with and without monitors are behaviorally equivalent)., Comment: Full version with appendices of our RV'23 paper

Cellular-Vehicle-to-Everything (C-V2X) is currently at the forefront of the digital transformation of our society. By enabling vehicles to communicate with each other and with the traffic environment using cellular networks, we redefine transportation, improving road safety and transportation services, increasing efficiency of vehicular traffic flows, and reducing environmental impact. To effectively facilitate the provisioning of Cellular Vehicular-to-Network (C-V2N) services, we tackle the interdependent problems of service task placement and scaling of edge resources. Specifically, we formulate the joint problem and prove that it is not computationally tractable. To address its complexity we introduce a Deep Hybrid Policy Gradient (DHPG), a Deep Reinforcement Learning (DRL) approach for hybrid action spaces.The performance of DHPG is evaluated against several state-of-the-art (SoA) solutions through simulations employing a real-world C-V2N traffic dataset. The results demonstrate that DHPG outperforms SoA solutions in maintaining C-V2N service latency below the preset delay threshold, while simultaneously optimizing the utilization of computing resources. Finally, time complexity analysis is conducted to verify that the proposed approach can support real-time C-V2N services., Comment: This paper has been submitted to IEEE Transactions on Mobile Computing

Last years we have witnessed the uprising of location based applications, which depend on the devices ability to accurately obtain their position. IEEE 802.11, foretelling the need for such applications, started the IEEE 802.11az work on Next Generation Positioning. Although this standard provides positioning enhancements for sub-6GHz and mmWave bands, high accuracy in the order of centimeters can only be obtained in the latter band, thanks to the beamforming information available at mmWave operation. This work presents a detailed analysis on the new techniques provided by IEEE 802.11az for enhanced secured positioning in the mmWave band, assessing them through experimentation., Comment: 8 pages, 6 figures, magazine submission

We present a simple and quick method to approximate network centrality indexes. Our approach, called QuickCent, is inspired by so-called fast and frugal heuristics, which are heuristics initially proposed to model some human decision and inference processes. The centrality index that we estimate is the harmonic centrality, which is a measure based on shortest-path distances, so infeasible to compute on large networks. We compare QuickCent with known machine learning algorithms on synthetic data generated with preferential attachment, and some empirical networks. Our experiments show that QuickCent is able to make estimates that are competitive in accuracy with the best alternative methods tested, either on synthetic scale-free networks or empirical networks. QuickCent has the feature of achieving low error variance estimates, even with a small training set. Moreover, QuickCent is comparable in efficiency -- accuracy and time cost -- to those produced by more complex methods. We discuss and provide some insight into how QuickCent exploits the fact that in some networks, such as those generated by preferential attachment, local density measures such as the in-degree, can be a proxy for the size of the network region to which a node has access, opening up the possibility of approximating centrality indices based on size such as the harmonic centrality. Our initial results show that simple heuristics and biologically inspired computational methods are a promising line of research in the context of network measure estimations.

DevOps responds the growing need of companies to streamline the software development process and, thus, has experienced widespread adoption in the past years. However, the successful adoption of DevOps requires companies to address significant cultural and organizational changes. Understanding the organizational structure and characteristics of teams adopting DevOps is key, and comprehending the existing theories and representations of team taxonomies is critical to guide companies in a more systematic and structured DevOps adoption process. As there was no unified theory to explain the different topologies of DevOps teams, in previous work, we built a theory to represent the organizational structure and characteristics of teams adopting DevOps, harmonizing the existing knowledge. In this paper, we expand the theory-building in the context of DevOps Team Taxonomies. Our main contributions are presenting and executing the Operationalization and Testing phases for a continuously evolving theory on DevOps team structures. We operationalize the constructs and propositions that make up our theory to generate empirically testable hypotheses to confirm or disconfirm the theory. Specifically, we focus on the research operation side of the theory-research cycle: identifying propositions, deriving empirical indicators from constructs, establishing testable hypotheses, and testing them. We performed the operationalization and testing of the DevOps Team Taxonomies Theory, which resulted in an empirically verified and trustworthy theory. Our theory has 28 propositions representing this model that map properties to the constructs of our theory. The operationalization generated 34 testable hypotheses, and we thoroughly tested 11 of them. The testing has proved the effectiveness of the theoretical framework, while the operationalization of the constructs has enhanced the initial framework., Comment: Submitted to Empirical Software Engineering

The fifth generation (5G) of wireless networks is set out to meet the stringent requirements of vehicular use cases. Edge computing resources can aid in this direction by moving processing closer to end-users, reducing latency. However, given the stochastic nature of traffic loads and availability of physical resources, appropriate auto-scaling mechanisms need to be employed to support cost-efficient and performant services. To this end, we employ Deep Reinforcement Learning (DRL) for vertical scaling in Edge computing to support vehicular-to-network communications. We address the problem using Deep Deterministic Policy Gradient (DDPG). As DDPG is a model-free off-policy algorithm for learning continuous actions, we introduce a discretization approach to support discrete scaling actions. Thus we address scalability problems inherent to high-dimensional discrete action spaces. Employing a real-world vehicular trace data set, we show that DDPG outperforms existing solutions, reducing (at minimum) the average number of active CPUs by 23% while increasing the long-term reward by 24%., Comment: Accepted at the 36th IEEE/IFIP Network Operations and Management Symposium (NOMS 2023)

Session types are a type-based approach to the verification of message-passing programs. They specify communication structures essential for program correctness; a session type says what and when should be exchanged through a channel. Central to session-typed languages are sequencing constructs in types and processes that explicitly specify the order of actions in a protocol. In this paper we study session types without sequencing. The resulting framework of minimal session types is arguably the simplest form of session types one could conceive. In the context of a core process calculus with sessions and higher-order concurrency (abstraction-passing), we establish two main technical results. First, we prove that every process $P$ typable with standard session types can be compiled down into a process $\mathcal{D}(P)$ typable with minimal session types. Second, we prove that $P$ and $\mathcal{D}(P)$ are behaviorally equivalent. These results indicate that having sequencing constructs in processes and session types is convenient but redundant: only sequentiality in processes is truly indispensable, as it can correctly codify sequentiality in types. Our developments draw inspiration from work by Parrow on behavior-preserving decompositions of untyped processes. By casting Parrow's results in the realm of typed processes, our developments reveal a conceptually simple formulation of session types and a principled avenue to the integration of session types into programming languages without sequencing in types., Comment: arXiv admin note: text overlap with arXiv:1906.03836

The future Internet promises to support time-sensitive services that require ultra low latencies and reliabilities of 99.99%. Recent advances in cellular and WiFi connections enhance the network to meet high reliability and ultra low latencies. However, the aforementioned services require that the server processing time ensures low latencies with high reliability, otherwise the end-to-end performance is not met. To that end, in this paper we use queuing theory to model the sojourn time distribution for Ultra-Reliable Low Latency Communication services of M/M/R-JSQ-PS systems: Markovian queues with R CPU servers following a join shortest queue processor-sharing discipline (for example Linux systems). We develop open-source simulation software, and develop and compare six analytical approximations for the sojourn time distribution. The proposed approximations yield Wasserstein distances below 2 time units, and upon medium loads incur into errors of less than 1.78 time units (e.g., milliseconds) for the 99.99th percentile sojourn time. Moreover, the proposed sojourn time approximations are stable regardless the number of CPUs and stay close to the simulations., Comment: 14 Pages, 10 figures, submitted to Elsevier European Journal of Operational Research

Recent trends in robotic services propose offloading robot functionalities to the Edge to meet the strict latency requirements of networked robotics. However, the Edge is typically an expensive resource and sometimes the Cloud is also an option, thus, decreasing the cost. Following this idea, we propose Don't Let Me Down! (DLMD), an algorithm that promotes offloading robot functions to the Cloud when possible to minimize the consumption of Edge resources. Additionally, DLMD takes the appropriate migration, traffic steering, and radio handover decisions to meet robotic service requirements as strict latency constraints. In the paper, we formulate the optimization problem that DLMD aims to solve, compare DLMD performance against state of art, and perform stress tests to assess DLMD performance in small & large networks. Results show that DLMD (i) always finds solutions in less than 30ms; (ii) is optimal in a local warehousing use case, and (iii) consumes only 5% of the Edge resources upon network stress., Comment: 5 Pages, 6 figures, submitted to 2023 IEEE 9th International Conference on Network Softwarization (NetSoft)