Sharif Digital Repository

Transferring large models in federated learning (FL) networks is often hindered by clients' limited bandwidth. We propose $ extsf {FedAA}$ , an FL algorithm which achieves fast convergence by exploiting the regularized Anderson acceleration (AA) on the global level. First, we demonstrate that FL can benefit from acceleration methods in numerical analysis. Second, $ extsf {FedAA}$ improves the convergence rate for quadratic losses and improves the empirical performance for smooth and strongly convex objectives, compared to FedAvg, an FL algorithm using gradient descent (GD) local updates. Experimental results demonstrate that employing AA can significantly improve the performance of FedAvg,... 

There is growing interest in applying distributed machine learning to edge computing, forming federated edge learning. Federated edge learning faces non-i.i.d. and heterogeneous data, and the communication between edge workers, possibly through distant locations and with unstable wireless networks, is more costly than their local computational overhead. In this work, we propose ${{sf DONE}}$DONE, a distributed approximate Newton-type algorithm with fast convergence rate for communication-efficient federated edge learning. First, with strongly convex and smooth loss functions, ${{sf DONE}}$DONE approximates the Newton direction in a distributed manner using the classical Richardson iteration... 

Often, recommendation systems employ continuous training, leading to a self-feedback loop bias in which the system becomes biased toward its previous recommendations. Recent studies have attempted to mitigate this bias by collecting small amounts of unbiased data. While these studies have successfully developed less biased models, they ignore the crucial fact that the recommendations generated by the model serve as the training data for subsequent training sessions. To address this issue, we propose a framework that learns an unbiased estimator using a small amount of uniformly collected data and focuses on generating improved training data for subsequent training iterations. To accomplish... 

We propose the first method that realizes the Laplace mechanism exactly (i.e., a Laplace noise is added to the data) that requires only a finite amount of communication (whereas the original Laplace mechanism requires the transmission of a real number) while guaranteeing privacy against the server and database. Our mechanism can serve as a drop-in replacement for local or centralized differential privacy applications where the Laplace mechanism is used. Our mechanism is constructed using a random quantization technique. Unlike the simple and prevalent Laplace-mechanism-then-quantize approach, the quantization in our mechanism does not result in any distortion or degradation of utility.... 

Natural convection in a porous enclosure in the presence of thermal dispersion is investigated. The Fourier–Galerkin (FG) spectral element method is adapted to solve the coupled equations of Darcy’s flow and heat transfer with a full velocity-dependent dispersion tensor, employing the stream function formulation. A sound implementation of the FG method is developed to obtain accurate solutions within affordable computational costs. In the spectral space, the stream function is expressed analytically in terms of temperature, and the spectral system is solved using temperature as the primary unknown. The FG method is compared to finite element solutions obtained using an in-house code... 

Non-pumping reactive wells (NPRWs) filled by zero-valent iron (ZVI) can be utilized for the remediation of groundwater contamination of deep aquifers. The efficiency of NPRWs mainly depends on the hydraulic contact time (HCT) of the pollutant with the reactive materials, the extent of the well capture zone (Wcz), and the relative hydraulic conductivity of aquifer and reactive material (Kr). We investigated nitrate removal from groundwater using NPRWs filled by ZVI (in nano and micro scales) and examined the effect of NPRWs orientations (i.e. vertical, slanted, and horizontal) on HCT and Wcz. The dependence of HCT on Wcz for different Kr values was derived theoretically for a homogeneous and... 

We revisit and shed light on the millennium-old hydrogeology textbook The Extraction of Hidden Waters by the Persian mathematician and engineer Karaji. Despite the nature of the understanding and conceptualization of the world by the people of that time, ground-breaking ideas and descriptions of hydrological and hydrogeological perceptions such as components of hydrological cycle, groundwater quality and even driving factors for groundwater flow were presented in the book. Although some of these ideas may have been presented elsewhere, to the best of our knowledge, this is the first time that a whole book was focused on different aspects of hydrology and hydrogeology. More importantly, we... 

Solar hydrogen evolution from water is a necessary step to overcome the challenges of rising energy demand and associated environmental concerns. Low-cost photocatalytic architectures based on polymeric light absorbers coupled to highly efficient molecular catalysts might represent an attractive platform to address this issue. However, to-date, our mechanistic knowledge of these systems is still largely underdeveloped. In this study, a molecular molybdenum sulfide hydrogen evolving catalyst, [Mo3S13]2-, is loaded onto polymeric carbon nitride (CNx) photoabsorber by impregnation. The resulting composite shows enhanced photocatalytic activity for hydrogen evolution compared to pristine CNx... 

In this research work, Friction Theory and Free Volume Theory are applied to live oil characterized based on SARA TEST for viscosity modeling and make a new model in combination with two equation of state (PR and PC-SAFT). Parameters for pseudo-components are obtained by tuning the viscosity at atmospheric pressure and temperatures of 10, 20, and 40 ℃. A new fitting approach is suggested where the number of fitting parameters is 17 and 12 for FT and FVT model, respectively. These parameters are tuned using the Genetic Algorithm and Particle Swarm Optimization and make eight new models. The results show that PC-SAFT provides viscosity predictions for all models with less deviation from... 

KeeLoq remote keyless entry systems are widely used for access control purposes such as garage openers or car door systems. We present the first successful differential power analysis attacks on numerous commercially available products employing KeeLoq code hopping. Our new techniques combine side-channel cryptanalysis with specific properties of the KeeLoq algorithm. They allow for efficiently revealing both the secret key of a remote transmitter and the manufacturer key stored in a receiver. As a result, a remote control can be cloned from only ten power traces, allowing for a practical key recovery in few minutes. After extracting the manufacturer key once, with similar techniques, we... 

In this paper a finite volume (FV) numerical method is implemented to solve a Biot consolidation model with discontinuous coefficients. Our studies show that the FV scheme leads to a locally mass conservative approach which removes pressure oscillations especially along the interface between materials with different properties and yields higher accuracy for the flow and mechanics parameters. Then this numerical discretization is utilized to investigate different sequential strategies with various degrees of coupling including: iteratively, explicitly and loosely coupled methods. A comprehensive study is performed on the stability, accuracy and rate of convergence of all of these sequential... 

Wax precipitation may occur in production or transportation of crude oil form field which is a serious problem in petroleum industry. Flow assurance issues concerning wax precipitation make it necessary to develop a precise thermodynamic model to predict the wax appearance temperature and amount of precipitation at different conditions. In this work a new procedure has been proposed to characterize crude oil based on the SARA test considering the wax and asphaltene as single pseudo components. Two scenarios have been investigated for the survey of the crude oil characterization, with and without asphaltene pseudo component. Also, in this work, the Perturbed Chain form of the Statistical... 

As a way of obtaining useful information about the adversaries behavior with a low rate of false detection, honeypots have made significant advancements in the field of cybersecurity. They are also powerful in wasting the adversaries resources and attracting their attention from other critical assets in the network. A deceptive network with multiple honeypots is called a honeynet. The honeypots in a honeynet aim to cooperate in order to increase their deception power. Professional adversaries utilize strong detection mechanisms to discover the existence of the honeypots in a network. When an adversary finds that a deception mechanism is deployed, it may change their behavior and cause... 

Network slicing in 5G networks can be modeled as a Virtual Network Embedding (VNE) problem, wherein the slice requests must be efficiently mapped on the core network. This process faces two major challenges: covering the maximum number of requests and providing slice isolation. Slice isolation is a mechanism for protecting the slices against Distributed Denial of Service (DDoS) attacks. To overcome these two challenges, we have proposed a novel actor-critic Reinforcement Learning (RL) model, called Slice Isolation-based Reinforcement Learning (SIRL), using five optimal graph features to create the problem environment, the form of which is changed based on a ranking scheme. The ranking... 

Network slicing within 5G networks encounters two significant challenges: catering to a maximum number of requests while ensuring slice isolation. To address these challenges, we present an innovative actor-critic Reinforcement Learning (RL) model named 'Slice Isolation based on RL' (SIRL). This model employs five optimal graph features to construct the problem environment, the structure of which is adapted using a ranking scheme. This scheme effectively reduces feature dimensionality and enhances learning performance. SIRL was assessed through a comparative analysis with nine state-of-the-art RL models, utilizing four evaluation metrics. The average results demonstrate that SIRL outperforms... 

5G slices are susceptible to indirect Distributed Denial of Service (DDoS) attacks, where overwhelming traffic directed to one slice can also disrupt other slices sharing the same infrastructure Many current mitigation methods rely on a detection phase, which may not be effective against unknown or sophisticated attacks. Moving Target Defense (MTD) is a security mechanism that invalidates the adversary's collected information, and it can be deployed without the detection phase. In this paper, we propose a Slice Mutation technique based on Reinforcement Learning (SMRL) that reduces the impact of DDoS attacks on 5G slices while keeping the number of allocated slices acceptable. SMRL proposes a... 

Understanding controls of P movement through watersheds are essential for improved landscape management in intensively managed regions. Here, we analyze observational data from 104 gaged river sites and 176 nongaged river sites within agriculturally dominated watersheds of Minnesota, USA, to understand the role of landscape features, land use practices, climate variability, and biogeochemical processes in total, dissolved and particulate P dynamics at daily to annual scales. Our analyses demonstrate that factors mediating P concentration-discharge relationships varied greatly across watersheds and included near-channel sediment sources, lake and wetland interception, assimilation by algal P,... 

We present and analyse light curves of four transits of the Southern hemisphere extrasolar planetary system WASP-4, obtained with a telescope defocused so the radius of each point spread function was 17 arcsec (44 pixels). This approach minimizes both random and systematic errors, allowing us to achieve scatters of between 0.60 and 0.88 mmag per observation over complete transit events. The light curves are augmented by published observations and analysed using the jktebop code. The results of this process are combined with theoretical stellar model predictions to derive the physical properties of the WASP-4 system. We find that the mass and radius of the planet are M b = 1.289 +0.090 -0.090... 

Real world models of seawater intrusion (SWI) require high computational efforts. This creates computational difficulties for the uncertainty propagation (UP) analysis of these models due the need for repeated numerical simulations in order to adequately capture the underlying statistics that describe the uncertainty in model outputs. Moreover, despite the obvious advantages of moment-independent global sensitivity analysis (SA) methods, these methods have rarely been employed for SWI and other complex groundwater models. The reason is that moment-independent global SA methods involve repeated UP analysis which further becomes computationally demanding. This study proposes the use of...