arXiv subjects

Leandros Tassiulas

Publications and source records attributed to Leandros Tassiulas.

At least 55 records · Page 3Linked to original sources

arXiv2025

An Item is Worth a Prompt: Versatile Image Editing with Disentangled Control

Building on the success of text-to-image diffusion models (DPMs), image editing is an important application to enable human interaction with AI-generated content. Among various editing methods, editing within the prompt space gains more attention due to its capacity and simplicity of controlling semantics. However, since diffusion models are commonly pretrained on descriptive text captions, direct editing of words in text prompts usually leads to completely different generated images, violating the requirements for image editing. On the other hand, existing editing methods usually consider introducing spatial masks to preserve the identity of unedited regions, which are usually ignored by DPMs and therefore lead to inharmonic editing results. Targeting these two challenges, in this work, we propose to disentangle the comprehensive image-prompt interaction into several item-prompt interactions, with each item linked to a special learned prompt. The resulting framework, named D-Edit, is based on pretrained diffusion models with cross-attention layers disentangled and adopts a two-step optimization to build item-prompt associations. Versatile image editing can then be applied to specific items by manipulating the corresponding prompts. We demonstrate state-of-the-art results in four types of editing operations including image-based, text-based, mask-based editing, and item removal, covering most types of editing applications, all within a single unified framework. Notably, D-Edit is the first framework that can (1) achieve item editing through mask editing and (2) combine image and text-based editing. We demonstrate the quality and versatility of the editing results for a diverse collection of images through both qualitative and quantitative evaluations.

cs.CV↗

arXiv2024

Mobility as a Resource (MaaR) for resilient human-centric automation: a vision paper

With technological advances, mobility has been moving from a product (i.e., traditional modes and vehicles), to a service (i.e., Mobility as a Service, MaaS). However, as observed in other fields (e.g. cloud computing resource management) we argue that mobility will evolve from a service to a resource (i.e., Mobility as a Resource, MaaR). Further, due to increasing scarcity of shared mobility spaces across traditional and emerging modes, the transition must be viewed within the critical need for ethical and equitable solutions for the traveling public (i.e., research is needed to avoid hyper-market driven outcomes for society). The evolution of mobility into a resource requires novel conceptual frameworks, technologies, processes and perspectives of analysis. A key component of the future MaaR system is the technological capacity to observe, allocate and manage (in real-time) the smallest envisionable units of mobility (i.e., atomic units of mobility capacity) while providing prioritized attention to human movement and ethical metrics related to access, consumption and impact. To facilitate research into the envisioned future system, this paper proposes initial frameworks which synthesize and advance methodologies relating to highly dynamic capacity reservation systems. Future research requires synthesis across transport network management, demand behavior, mixed-mode usage, and equitable mobility.

eess.SY↗

arXiv2024

From Similarity to Superiority: Channel Clustering for Time Series Forecasting

Time series forecasting has attracted significant attention in recent decades. Previous studies have demonstrated that the Channel-Independent (CI) strategy improves forecasting performance by treating different channels individually, while it leads to poor generalization on unseen instances and ignores potentially necessary interactions between channels. Conversely, the Channel-Dependent (CD) strategy mixes all channels with even irrelevant and indiscriminate information, which, however, results in oversmoothing issues and limits forecasting accuracy. There is a lack of channel strategy that effectively balances individual channel treatment for improved forecasting performance without overlooking essential interactions between channels. Motivated by our observation of a correlation between the time series model's performance boost against channel mixing and the intrinsic similarity on a pair of channels, we developed a novel and adaptable Channel Clustering Module (CCM). CCM dynamically groups channels characterized by intrinsic similarities and leverages cluster information instead of individual channel identities, combining the best of CD and CI worlds. Extensive experiments on real-world datasets demonstrate that CCM can (1) boost the performance of CI and CD models by an average margin of 2.4% and 7.2% on long-term and short-term forecasting, respectively; (2) enable zero-shot forecasting with mainstream time series forecasting models; (3) uncover intrinsic time series patterns among channels and improve interpretability of complex time series models.

cs.LG↗

arXiv2024

SplitLLM: Collaborative Inference of LLMs for Model Placement and Throughput Optimization

Large language models (LLMs) have been a disruptive innovation in recent years, and they play a crucial role in our daily lives due to their ability to understand and generate human-like text. Their capabilities include natural language understanding, information retrieval and search, translation, chatbots, virtual assistance, and many more. However, it is well known that LLMs are massive in terms of the number of parameters. Additionally, the self-attention mechanism in the underlying architecture of LLMs, Transformers, has quadratic complexity in terms of both computation and memory with respect to the input sequence length. For these reasons, LLM inference is resource-intensive, and thus, the throughput of LLM inference is limited, especially for the longer sequences. In this report, we design a collaborative inference architecture between a server and its clients to alleviate the throughput limit. In this design, we consider the available resources on both sides, i.e., the computation and communication costs. We develop a dynamic programming-based algorithm to optimally allocate computation between the server and the client device to increase the server throughput, while not violating the service level agreement (SLA). We show in the experiments that we are able to efficiently distribute the workload allowing for roughly 1/3 reduction in the server workload, while achieving 19 percent improvement over a greedy method. As a result, we are able to demonstrate that, in an environment with different types of LLM inference requests, the throughput of the server is improved.

cs.DC↗

arXiv2024

Throughput-Optimal Scheduling via Rate Learning

We study the problem of designing scheduling policies for communication networks. This problem is often addressed with max-weight-type approaches since they are throughput-optimal. However, max-weight policies make scheduling decisions based on the network congestion, which can be sometimes unnecessarily restrictive. In this paper, we present a ``schedule as you learn'' (SYL) approach, where we learn an average rate, and then select schedules that generate such a rate in expectation. This approach is interesting because scheduling decisions do not depend on the size of the queue backlogs, and so it provides increased flexibility to select schedules based on other criteria or rules, such as serving high-priority queues. We illustrate the results with numerical experiments for a cross-bar switch and show that, compared to max-weight, SYL can achieve lower latency to certain flows without compromising throughput optimality.

cs.NI↗

arXiv2024

Over-the-Air Federated Learning via Weighted Aggregation

This paper introduces a new federated learning scheme that leverages over-the-air computation. A novel feature of this scheme is the proposal to employ adaptive weights during aggregation, a facet treated as predefined in other over-the-air schemes. This can mitigate the impact of wireless channel conditions on learning performance, without needing channel state information at transmitter side (CSIT). We provide a mathematical methodology to derive the convergence bound for the proposed scheme in the context of computational heterogeneity and general loss functions, supplemented with design insights. Accordingly, we propose aggregation cost metrics and efficient algorithms to find optimized weights for the aggregation. Finally, through numerical experiments, we validate the effectiveness of the proposed scheme. Even with the challenges posed by channel conditions and device heterogeneity, the proposed scheme surpasses other over-the-air strategies by an accuracy improvement of 15% over the scheme using CSIT and 30% compared to the one without CSIT.

cs.IT↗

arXiv2024

A Brief Introduction to Quantum Network Control

Quantum networking is an emerging area with the potential to transform information processing and communications. In this paper, we present a brief introduction to quantum network control, an area in quantum networking dedicated to designing algorithms for distributing entanglement (i.e., entangled qubits). We start by explaining what qubits and entanglement are and how they furnish quantum network control operations such as entanglement swapping and teleportation. With those operations, we present a model for distributing entanglement in a multi-hop quantum network to enable applications such as quantum key distribution and distributed quantum computing. We conclude the paper by presenting open research problems in the field, including the characterization of the quantum network capacity region and the design of throughput-optimal policies.

math.OC↗

arXiv2024

On the Capacity of the Quantum Switch with and without Entanglement Decoherence

This paper studies the capacity of the quantum switch for two decoherence models: when link-level entanglements last (i) for a time slot, or (ii) until they are used to serve a request (i.e., there is no decoherence). The two models are important as they set lower and upper bounds on the capacity region for any other decoherence model. The paper's contributions are to characterize the switch capacity region for both decoherence models and to propose throughput-optimal policies based on gradient descent.

math.OC↗

arXiv2024

Compiler for Distributed Quantum Computing: a Reinforcement Learning Approach

The practical realization of quantum programs that require large-scale qubit systems is hindered by current technological limitations. Distributed Quantum Computing (DQC) presents a viable path to scalability by interconnecting multiple Quantum Processing Units (QPUs) through quantum links, facilitating the distributed execution of quantum circuits. In DQC, EPR pairs are generated and shared between distant QPUs, which enables quantum teleportation and facilitates the seamless execution of circuits. A primary obstacle in DQC is the efficient mapping and routing of logical qubits to physical qubits across different QPUs, necessitating sophisticated strategies to overcome hardware constraints and optimize communication. We introduce a novel compiler that, unlike existing approaches, prioritizes reducing the expected execution time by jointly managing the generation and routing of EPR pairs, scheduling remote operations, and injecting SWAP gates to facilitate the execution of local gates. We present a real-time, adaptive approach to compiler design, accounting for the stochastic nature of entanglement generation and the operational demands of quantum circuits. Our contributions are twofold: (i) we model the optimal compiler for DQC using a Markov Decision Process (MDP) formulation, establishing the existence of an optimal algorithm, and (ii) we introduce a constrained Reinforcement Learning (RL) method to approximate this optimal compiler, tailored to the complexities of DQC environments. Our simulations demonstrate that Double Deep Q-Networks (DDQNs) are effective in learning policies that minimize the depth of the compiled circuit, leading to a lower expected execution time and likelihood of successful operation before qubits decohere.

quant-ph↗

arXiv2024

Adaptive Heterogeneous Client Sampling for Federated Learning over Wireless Networks

Federated learning (FL) algorithms usually sample a fraction of clients in each round (partial participation) when the number of participants is large and the server's communication bandwidth is limited. Recent works on the convergence analysis of FL have focused on unbiased client sampling, e.g., sampling uniformly at random, which suffers from slow wall-clock time for convergence due to high degrees of system heterogeneity and statistical heterogeneity. This paper aims to design an adaptive client sampling algorithm for FL over wireless networks that tackles both system and statistical heterogeneity to minimize the wall-clock convergence time. We obtain a new tractable convergence bound for FL algorithms with arbitrary client sampling probability. Based on the bound, we analytically establish the relationship between the total learning time and sampling probability with an adaptive bandwidth allocation scheme, which results in a non-convex optimization problem. We design an efficient algorithm for learning the unknown parameters in the convergence bound and develop a low-complexity algorithm to approximately solve the non-convex problem. Our solution reveals the impact of system and statistical heterogeneity parameters on the optimal client sampling design. Moreover, our solution shows that as the number of sampled clients increases, the total convergence time first decreases and then increases because a larger sampling number reduces the number of rounds for convergence but results in a longer expected time per-round due to limited wireless bandwidth. Experimental results from both hardware prototype and simulation demonstrate that our proposed sampling scheme significantly reduces the convergence time compared to several baseline sampling schemes.

cs.DC↗

arXiv2024

Machine Learning on Blockchain Data: A Systematic Mapping Study

Context: Blockchain technology has drawn growing attention in the literature and in practice. Blockchain technology generates considerable amounts of data and has thus been a topic of interest for Machine Learning (ML). Objective: The objective of this paper is to provide a comprehensive review of the state of the art on machine learning applied to blockchain data. This work aims to systematically identify, analyze, and classify the literature on ML applied to blockchain data. This will allow us to discover the fields where more effort should be placed in future research. Method: A systematic mapping study has been conducted to identify the relevant literature. Ultimately, 159 articles were selected and classified according to various dimensions, specifically, the domain use case, the blockchain, the data, and the machine learning models. Results: The majority of the papers (49.7%) fall within the Anomaly use case. Bitcoin (47.2%) was the blockchain that drew the most attention. A dataset consisting of more than 1.000.000 data points was used by 31.4% of the papers. And Classification (46.5%) was the ML task most applied to blockchain data. Conclusion: The results confirm that ML applied to blockchain data is a relevant and a growing topic of interest both in the literature and in practice. Nevertheless, some open challenges and gaps remain, which can lead to future research directions. Specifically, we identify novel machine learning algorithms, the lack of a standardization framework, blockchain scalability issues and cross-chain interactions as areas worth exploring in the future.

cs.CR↗

arXiv2024

Cyber-Twin: Digital Twin-boosted Autonomous Attack Detection for Vehicular Ad-Hoc Networks

The rapid evolution of Vehicular Ad-hoc NETworks (VANETs) has ushered in a transformative era for intelligent transportation systems (ITS), significantly enhancing road safety and vehicular communication. However, the intricate and dynamic nature of VANETs presents formidable challenges, particularly in vehicle-to-infrastructure (V2I) communications. Roadside Units (RSUs), integral components of VANETs, are increasingly susceptible to cyberattacks, such as jamming and distributed denial of service (DDoS) attacks. These vulnerabilities pose grave risks to road safety, potentially leading to traffic congestion and vehicle malfunctions. Existing methods face difficulties in detecting dynamic attacks and integrating digital twin technology and artificial intelligence (AI) models to enhance VANET cybersecurity. Our study proposes a novel framework that combines digital twin technology with AI to enhance the security of RSUs in VANETs and address this gap. This framework enables real-time monitoring and efficient threat detection while also improving computational efficiency and reducing data transmission delay for increased energy efficiency and hardware durability. Our framework outperforms existing solutions in resource management and attack detection. It reduces RSU load and data transmission delay while achieving an optimal balance between resource consumption and high attack detection effectiveness. This highlights our commitment to secure and sustainable vehicular communication systems for smart cities.

cs.CR↗

arXiv2024

Efficient High-Resolution Time Series Classification via Attention Kronecker Decomposition

The high-resolution time series classification problem is essential due to the increasing availability of detailed temporal data in various domains. To tackle this challenge effectively, it is imperative that the state-of-the-art attention model is scalable to accommodate the growing sequence lengths typically encountered in high-resolution time series data, while also demonstrating robustness in handling the inherent noise prevalent in such datasets. To address this, we propose to hierarchically encode the long time series into multiple levels based on the interaction ranges. By capturing relationships at different levels, we can build more robust, expressive, and efficient models that are capable of capturing both short-term fluctuations and long-term trends in the data. We then propose a new time series transformer backbone (KronTime) by introducing Kronecker-decomposed attention to process such multi-level time series, which sequentially calculates attention from the lower level to the upper level. Experiments on four long time series datasets demonstrate superior classification results with improved efficiency compared to baseline methods.

cs.LG↗

arXiv2024

Adaptive Compression-Aware Split Learning and Inference for Enhanced Network Efficiency

The growing number of AI-driven applications in mobile devices has led to solutions that integrate deep learning models with the available edge-cloud resources. Due to multiple benefits such as reduction in on-device energy consumption, improved latency, improved network usage, and certain privacy improvements, split learning, where deep learning models are split away from the mobile device and computed in a distributed manner, has become an extensively explored topic. Incorporating compression-aware methods (where learning adapts to compression level of the communicated data) has made split learning even more advantageous. This method could even offer a viable alternative to traditional methods, such as federated learning techniques. In this work, we develop an adaptive compression-aware split learning method ('deprune') to improve and train deep learning models so that they are much more network-efficient, which would make them ideal to deploy in weaker devices with the help of edge-cloud resources. This method is also extended ('prune') to very quickly train deep learning models through a transfer learning approach, which trades off little accuracy for much more network-efficient inference abilities. We show that the 'deprune' method can reduce network usage by 4x when compared with a split-learning approach (that does not use our method) without loss of accuracy, while also improving accuracy over compression-aware split-learning by 4 percent. Lastly, we show that the 'prune' method can reduce the training time for certain models by up to 6x without affecting the accuracy when compared against a compression-aware split-learning approach.

cs.NI↗

arXiv2024

Constrained Reinforcement Learning for Adaptive Controller Synchronization in Distributed SDN

In software-defined networking (SDN), the implementation of distributed SDN controllers, with each controller responsible for managing a specific sub-network or domain, plays a critical role in achieving a balance between centralized control, scalability, reliability, and network efficiency. These controllers must be synchronized to maintain a logically centralized view of the entire network. While there are various approaches for synchronizing distributed SDN controllers, most tend to prioritize goals such as optimization of communication latency or load balancing, often neglecting to address both the aspects simultaneously. This limitation becomes particularly significant when considering applications like Augmented and Virtual Reality (AR/VR), which demand constrained network latencies and substantial computational resources. Additionally, many existing studies in this field predominantly rely on value-based reinforcement learning (RL) methods, overlooking the potential advantages offered by state-of-the-art policy-based RL algorithms. To bridge this gap, our work focuses on examining deep reinforcement learning (DRL) techniques, encompassing both value-based and policy-based methods, to guarantee an upper latency threshold for AR/VR task offloading within SDN environments, while selecting the most cost-effective servers for AR/VR task offloading. Our evaluation results indicate that while value-based methods excel in optimizing individual network metrics such as latency or load balancing, policy-based approaches exhibit greater robustness in adapting to sudden network changes or reconfiguration.

cs.NI↗

arXiv2023

Joint SDN Synchronization and Controller Placement in Wireless Networks using Deep Reinforcement Learning

Software Defined Networking has afforded numerous benefits to the network users but there are certain persisting issues with this technology, two of which are scalability and privacy. The natural solution to overcoming these limitations is a distributed SDN controller architecture where multiple controllers are deployed over the network, with each controller orchestrating a certain segment of the network. However, since the centralized control is the key attribute of SDN that allows it to be so beneficial, a centralized logical view of the network will have to be maintained by each of these controllers; this can be done through synchronization of the distributed controllers, where each controller communicates with the others to ensure that they remain informed about the entire network. There is however a network cost associated with constantly having to update each others about different aspects of the network, which will become a greater issue in dynamic wireless networks. To minimize this network cost, there is a need to consider not only when to get the update information from the neighboring controllers, but also where to dynamically place the controllers such that the network costs may be minimized. The placement should take into consideration both communication for synchronization among the distributed controllers and communication of the controllers with the network devices that they manage. In this work, we show that our multi-objective deep reinforcement learning-based method performs the best at achieving different application goals by developing policy for controller synchronization as well as placement, outperforming different other possible approaches, under a wide variety of network conditions.

cs.NI↗

arXiv2023

Age Optimum Sampling in Non-Stationary Environment

In this work, we consider a status update system with a sensor and a receiver. The status update information is sampled by the sensor and then forwarded to the receiver through a channel with non-stationary delay distribution. The data freshness at the receiver is quantified by the Age-of-Information (AoI). The goal is to design an online sampling strategy that can minimize the average AoI when the non-stationary delay distribution is unknown. Assuming that channel delay distribution may change over time, to minimize the average AoI, we propose a joint stochastic approximation and non-parametric change point detection algorithm that can: (1) learn the optimum update threshold when the delay distribution remains static; (2) detect the change in transmission delay distribution quickly and then restart the learning process. Simulation results show that the proposed algorithm can quickly detect the delay changes, and the average AoI obtained by the proposed policy converges to the minimum AoI.

cs.NI↗

arXiv2023

Optimization methods for the capacitated refueling station location problem with routing

The energy transition in transportation benefits from demand-based models to determine the optimal placement of refueling stations for alternative fuel vehicles such as battery electric trucks. A formulation known as the refueling station location problem with routing (RSLP-R) is concerned with minimizing the number of stations necessary to cover a set of origin-destination trips such that the transit time does not exceed a given threshold. In this paper we extend the RSLP-R by station capacities to limit the number of vehicles that can be refueled at individual stations. The solution to the capacitated RSLP-R (CRSLP-R) avoids congestion of refueling stations by satisfying capacity constraints. We devise two optimization methods to deal with the increased difficulty to solve the CRSLP-R. The first method extends a prior branch-and-cut approach and the second method is a branch-cut-and-price algorithm based on variables associated with feasible routes. We evaluate both our methods on instances from the literature as well as a newly constructed network and find that the relative performance of the algorithms depends on the strictness of the capacity constraints. Furthermore, we show some runtime improvements over prior work on uncapacitated instances.

math.OC↗