Search

Graphene is a unique platform for tunable opto-electronic applications thanks to its linear band dispersion, which allows electrical control of resonant light-matter interactions. Tuning the nonlinear optical response of graphene is possible both electrically and in an all-optical fashion, but each approach involves a trade-off between speed and modulation depth. Here, we combine lattice temperature, electron doping, and all-optical tuning of third-harmonic generation in a hBN-encapsulated graphene opto-electronic device and demonstrate up to 85% modulation depth along with gate-tunable ultrafast dynamics. These results arise from the dynamic changes in the transient electronic temperature combined with Pauli blocking induced by the out-of-equilibrium chemical potential. Our work provides a detailed description of the transient nonlinear optical and electronic response of graphene, which is crucial for the design of nanoscale and ultrafast optical modulators, detectors and frequency converters., Comment: 28 pages, 6 figures

This paper investigates a multiuser millimeter-wave (mmWave) uplink system in which each user is equipped with a multi-antenna fluid antenna system (FAS) while the base station (BS) has multiple fixed-position antennas. Our primary objective is to maximize the system capacity by optimizing the transmit covariance matrices and the antenna position vectors of the users jointly. To gain insights, we start by deriving upper bounds and approximations for the capacity. Then we delve into the capacity maximization problem. Beginning with the simple scenario of a single user equipped with a single-antenna FAS, we demonstrate that a closed-form optimal solution exists when there are only two propagation paths between the user and the BS. In the case where multiple propagation paths are present, a near-optimal solution can also be obtained through a one-dimensional search method. Expanding our focus to multiuser cases, in which users are equipped with either single- or multi-antenna FAS, we show that the original capacity maximization problems can be reformulated into distinct rank-one programmings. Then, we propose alternating optimization algorithms to deal with the transformed problems. Simulation results indicate that FAS can improve the capacity of the multiple access channel (MAC) greatly, and the proposed algorithms outperform all the benchmarks., Comment: 17 pages, 8 figures

Generating a pure spin current using electrons, which have degrees of freedom beyond spin, such as electric charge and valley index, presents challenges. In response, we propose a novel mechanism based on intervalley exciton dynamics in {\em arc-shaped} strained transition metal dichalcogenides (TMDs) to achieve the {\em exciton spin Hall effect} in an electrically insulating regime, without the need for an external electric field. The interplay between strain gradients and strain-induced pseudomagnetic fields results in a net Lorentz force on long-lived intervalley excitons in WSe$_2$, carrying non-zero spin angular momentum. This process generates an exciton-mediated pure spin Hall current, resulting in opposite-sign spin accumulations and local magnetization on the two sides of the single-layer arc-shaped TMD. We demonstrate that the magnetic field induced by spin accumulation, at approximately $\sim {\rm mT}$, can be detected using techniques such as superconducting quantum interference magnetometry or spatially-resolved magneto-optical Faraday and Kerr rotations.

Hardware-aware Neural Architecture Search (HW-NAS) is a technique used to automatically design the architecture of a neural network for a specific task and target hardware. However, evaluating the performance of candidate architectures is a key challenge in HW-NAS, as it requires significant computational resources. To address this challenge, we propose an efficient hardware-aware evolution-based NAS approach called HW-EvRSNAS. Our approach re-frames the neural architecture search problem as finding an architecture with performance similar to that of a reference model for a target hardware, while adhering to a cost constraint for that hardware. This is achieved through a representation similarity metric known as Representation Mutual Information (RMI) employed as a proxy performance evaluator. It measures the mutual information between the hidden layer representations of a reference model and those of sampled architectures using a single training batch. We also use a penalty term that penalizes the search process in proportion to how far an architecture's hardware cost is from the desired hardware cost threshold. This resulted in a significantly reduced search time compared to the literature that reached up to 8000x speedups resulting in lower CO2 emissions. The proposed approach is evaluated on two different search spaces while using lower computational resources. Furthermore, our approach is thoroughly examined on six different edge devices under various hardware cost constraints., Comment: WACV 2024

Few-shot learning or meta-learning leverages the data scarcity problem in machine learning. Traditionally, training data requires a multitude of samples and labeling for supervised learning. To address this issue, we propose a one-shot unsupervised meta-learning to learn the latent representation of the training samples. We use augmented samples as the query set during the training phase of the unsupervised meta-learning. A temperature-scaled cross-entropy loss is used in the inner loop of meta-learning to prevent overfitting during unsupervised learning. The learned parameters from this step are applied to the targeted supervised meta-learning in a transfer-learning fashion for initialization and fast adaptation with improved accuracy. The proposed method is model agnostic and can aid any meta-learning model to improve accuracy. We use model agnostic meta-learning (MAML) and relation network (RN) on Omniglot and mini-Imagenet datasets to demonstrate the performance of the proposed method. Furthermore, a meta-learning model with the proposed initialization can achieve satisfactory accuracy with significantly fewer training samples.

This paper proposes a set of novel optimization algorithms for solving a class of convex optimization problems with time-varying streaming cost function. We develop an approach to track the optimal solution with a bounded error. Unlike the existing results, our algorithm is executed only by using the first-order derivatives of the cost function which makes it computationally efficient for optimization with time-varying cost function. We compare our algorithms to the gradient descent algorithm and show why gradient descent is not an effective solution for optimization problems with time-varying cost. Several examples including solving a model predictive control problem cast as a convex optimization problem with a streaming time-varying cost function demonstrate our results.

This is the second paper in a series, which studies the likelihood that some globular clusters (GCs) of the Milky Way (MW) could have originated from a dwarf satellite galaxy (DSG). Using a large suite of three-body simulations we determine the present-day orbital properties of 154 GCs that could have escaped from 41 MW DSGs over the past $8\,\mathrm{Gyrs}$. For the MW we considered two sets of static and dynamic models which account for the sustained growth of the MW since its birth. We focus on the Magellanic Clouds and Sagittarius. We compare the apogalactic distance, eccentricity, and orbital inclination of the MW GCs with those of runaway GCs from DSGs, to constrain their possible ex-situ origin. We observe a positive correlation between a DSG mass and the dispersion of its runaway GCs in the orbital parameter space of ($R_\mathrm{ap}$, $e$). We provide tables of the identified MW GCs and their likely associated progenitors. In total, we find 29 (19%) MW GCs which could be kinematically associated with MW DSGs. We report, for the first time, 6 and 10 new associations with the Large Magellanic Cloud and the Sagittarius, respectively. For the Sagittarius we predict a concentration of runaway GCs at large apogalactic distances of $R_\mathrm{ap}\approx275-375\,\mathrm{kpc}$, $e\approx0.8$, and a relative inclination of $\Delta\theta\approx20^{\circ}$. So far, there has not been any observed GCs with such orbital elements. Complemented with photometric and spectroscopic observations, and cosmological simulations, the findings from the present study could conclusively settle the debate over the in-situ vs. ex-situ origin of the MW GCs., Comment: 20 pages, 16 figures, 5 tables, references amended. Accepted for publication in MNRAS

Learning new tasks accumulatively without forgetting remains a critical challenge in continual learning. Generative experience replay addresses this challenge by synthesizing pseudo-data points for past learned tasks and later replaying them for concurrent training along with the new tasks' data. Generative replay is the best strategy for continual learning under a strict class-incremental setting when certain constraints need to be met: (i) constant model size, (ii) no pre-training dataset, and (iii) no memory buffer for storing past tasks' data. Inspired by the biological nervous system mechanisms, we introduce a time-aware regularization method to dynamically fine-tune the three training objective terms used for generative replay: supervised learning, latent regularization, and data reconstruction. Experimental results on major benchmarks indicate that our method pushes the limit of brain-inspired continual learners under such strict settings, improves memory retention, and increases the average performance over continually arriving tasks.

This paper which is part of the New Faculty Highlights Invited Speaker Program of AAAI'23, serves as a comprehensive survey of my research in transfer learning by utilizing embedding spaces. The work reviewed in this paper specifically revolves around the inherent challenges associated with continual learning and limited availability of labeled data. By providing an overview of my past and ongoing contributions, this paper aims to present a holistic understanding of my research, paving the way for future explorations and advancements in the field. My research delves into the various settings of transfer learning, including, few-shot learning, zero-shot learning, continual learning, domain adaptation, and distributed learning. I hope this survey provides a forward-looking perspective for researchers who would like to focus on similar research directions., Comment: to appear in AI Magazine Winter 2023 Issue

This paper investigates the performance of vehicleto-vehicle (V2V) communications assisted by a reconfigurable intelligent surface (RIS) and a simultaneous transmitting and reflecting intelligent omni-surface (STAR-IOS) under nonorthogonal multiple access (NOMA) and orthogonal multiple access (OMA) schemes. In particular, we consider that the RIS is close to the transmitter vehicle while the STAR-IOS is near the receiver vehicles. In addition, we assume that the STAR-IOS exploits the energy-splitting (ES) protocol for communication and the fading channels between the RIS and STAR-IOS follow composite Fisher-Snedecor F distribution. Under such assumptions, we first use the central limit theorem (CLT) to derive the PDF and the CDF of equivalent channels at receiver vehicles, and then, we derive the closed-form expression of outage probability (OP) under NOMA/OMA scenarios. Additionally, by exploiting Jensen's inequality, we propose an upper bound of the ergodic capacity (EC), and then, we derive an analytical expression of the energy efficiency (EE) for both NOMA and OMA cases. Further, our analytical results, which are double-checked with the Monte-Carlo simulation, reveal that applying RIS/STAR-RIS in V2V communications can significantly improve the performance of intelligent transportation systems (ITS). Besides, the results indicate that considering the NOMA scheme provides better performance in terms of the OP, EC, and EE as compared with the OMA case for the considered V2V communication.

This letter investigates the application of the emerging fluid antenna (FA) technology in multiuser communication systems when side information (SI) is available at the transmitters. In particular, we consider a K-user dirty multiple access channel (DMAC) with non-causally known SI at the transmitters, where K users send independent messages to a common receiver with a FA capable of changing its location depending on the channel condition. By connecting Jakes' model to copula theory through Spearman's {\rho} rank correlation coefficient, we accurately describe the spatial correlation between the FA channels, and derive a closed-form expression for the outage probability (OP) under Fisher-Snedecor F fading. Numerical results illustrate how considering FA can improve the performance of multiuser communication systems in terms of the OP and also support a large number of users using only one FA at the common receiver in a few wavelengths of space.

This paper investigates the performance of a single-user fluid antenna system (FAS), by exploiting a class of elliptical copulas to describe the dependence structure amongst the fluid antenna positions (ports). By expressing the well-known Jakes' model in terms of the Gaussian copula, we consider two cases: (i) the general case, i.e., any arbitrary correlated fading distribution; and (ii) the specific case, i.e., correlated Nakagami-$m$ fading. For both scenarios, we first derive analytical expressions for the cumulative distribution function (CDF) and probability density function (PDF) of the equivalent channel in terms of multivariate normal distribution. Then we obtain the outage probability (OP) and the delay outage rate (DOR) to analyze the performance of FAS. By employing the popular rank correlation coefficients such as Spearman's $\rho$ and Kendall's $\tau$ , we measure the degree of dependency in correlated arbitrary fading channels and illustrate how the Gaussian copula can be accurately connected to Jakes' model in FAS. Our numerical results demonstrate that increasing the size of FAS provides lower OP and DOR, but the system performance saturates as the number of antenna ports increases. In addition, our results indicate that FAS provides better performance compared to conventional single-fixed antenna systems even when the size of fluid antenna is small.

We categorize all non-abelian nilpotent Lie superalgebras of dimension $(m|n)$, where $1\leq s(L)\leq 10$, and $s(L)$ is a non-negative integer defined by Nayak. Furthermore, we classify the structure of all Lie superalgebras of dimension at most five such that $\dim{L^2}=\dim\mathcal{M}(L)$., Comment: In this revised version, all typographical errors have been meticulously corrected, enhancing clarity and precision. Additionally, a previously mentioned theorem has been revised to incorporate the most recent information or insights, thereby improving its overall accuracy and readability

We study the spreading of viscous and viscoelastic drops on solid substrates with different wettability. In the early stages of spreading, we find that the viscoelastic drop spreads with faster and a different power law than the Newtonian drop (i.e. aqueous glycerine solution) for the same zero shear rate viscosity. We argue that the effect of viscoelasticity is only observable for experimental time scales in the order of the internal relaxation time of the polymer solution or longer times. Near the contact line, the effective viscosity is lower for the viscoelastic drop than for the Newtonian drop. Together with its shear rate dependency, this difference in effective viscosity can explain the different spreading dynamics. We support our experimental findings with a simple perturbation model that qualitatively agrees with our findings.

Thermal-aware workload distribution is a common approach in the literature for power consumption optimization in data centers. However, data centers also have other operational costs such as the cost of equipment maintenance and replacement. It has been shown that server reliability depends on frequency of their temperature variations, arising from workload transitions due to dynamic demands. In this work, we formulate a nonlinear optimization problem that considers the cost of workload transitions in addition to IT and cooling power consumption. To approximate the solution, we first linearize the problem; the result is a mixed integer programming problem. A modified heuristic is then proposed to approximate the solution of the linear problem. Finally, a Model Predictive Control (MPC) approach is integrated with the proposed heuristics for automatic workload reconfiguration when future demand is not known exactly, but predictions are available. Numerical results show that the proposed schemes are attractive in different settings.

We prove that every profinite $n$-ary group $(G, f)=\Gf$ has a unique Haar measure $m_p$ and further for every measurable subset $A\subseteq G$, we have $$ m_p(A)=m(A)=(n-1)m^{\ast}(A) $$ where $m$ and $m^{\ast}$ are the normalized Haar measures of the profinite groups $(G, \bullet)$ and the Post cover $G^{\ast}$, respectively.

Despite significant success of deep learning in object detection tasks, the standard training of deep neural networks requires access to a substantial quantity of annotated images across all classes. Data annotation is an arduous and time-consuming endeavor, particularly when dealing with infrequent objects. Few-shot object detection (FSOD) methods have emerged as a solution to the limitations of classic object detection approaches based on deep learning. FSOD methods demonstrate remarkable performance by achieving robust object detection using a significantly smaller amount of training data. A challenge for FSOD is that instances from novel classes that do not belong to the fixed set of training classes appear in the background and the base model may pick them up as potential objects. These objects behave similarly to label noise because they are classified as one of the training dataset classes, leading to FSOD performance degradation. We develop a semi-supervised algorithm to detect and then utilize these unlabeled novel objects as positive samples during the FSOD training stage to improve FSOD performance. Specifically, we develop a hierarchical ternary classification region proposal network (HTRPN) to localize the potential unlabeled novel objects and assign them new objectness labels to distinguish these objects from the base training dataset classes. Our improved hierarchical sampling strategy for the region proposal network (RPN) also boosts the perception ability of the object detection model for large objects. We test our approach and COCO and PASCAL VOC baselines that are commonly used in FSOD literature. Our experimental results indicate that our method is effective and outperforms the existing state-of-the-art (SOTA) FSOD methods. Our implementation is provided as a supplement to support reproducibility of the results., Comment: arXiv admin note: substantial text overlap with arXiv:2303.10422

We study the inflationary scenario in the Tsallis entropy-based cosmology. The Friedmann equations in this setup can be derived by using the first law of thermodynamics. To derive the relations of the power spectra of the scalar and tensor perturbations in this setup, we reconstruct an $f(R)$ gravity model which is thermodynamically equivalent to our model in the slow-roll approximation. In this way, we find the inflationary observables, including the scalar spectral index and the tensor-to-scalar ratio in our scenario. Then, we investigate two different potentials in our scenario, including the quadratic potential and the potential associated with the natural inflation in which the inflaton is an axion or a pseudo-Nambu-Goldstone boson. We examine their observational viability in light of the Planck 2018 CMB data. We show that although the results of these potentials are in tension with the observations in the standard inflationary setting, their consistency with the observations can be significantly improved within the setup of the Tsallis entropy-based inflation. Moreover, we place constraints on the parameters of the considered inflationary models by using the current observational data., Comment: 37 pages, 9 figures

5G non-public networks (NPNs) play a key role in enabling critical Industrial Internet of Things (IoT) applications in various vertical industries. Among other features, 5G NPNs enable novel operation models, where the roles and responsibilities for setting up and operating the network can be distributed among several stakeholders, i.e., among the public mobile network operators (MNOs), the industrial party who uses the 5G NPN services and 3rd parties. This results in many theoretically feasible operation models for 5G NPN, each with its own advantages and disadvantages. We investigate the resulting operation models and identify a set of nine prime models taking into account today's practical considerations. Additionally, we define a framework to qualitatively analyze the operation models and use it to evaluate and compare the identified operation models.

Deploying deep learning neural networks on edge devices, to accomplish task specific objectives in the real-world, requires a reduction in their memory footprint, power consumption, and latency. This can be realized via efficient model compression. Disentangled latent representations produced by variational autoencoder (VAE) networks are a promising approach for achieving model compression because they mainly retain task-specific information, discarding useless information for the task at hand. We make use of the Beta-VAE framework combined with a standard criterion for pruning to investigate the impact of forcing the network to learn disentangled representations on the pruning process for the task of classification. In particular, we perform experiments on MNIST and CIFAR10 datasets, examine disentanglement challenges, and propose a path forward for future works., Comment: Presented in ISCS23

Estimation of the Average Treatment Effect (ATE) is often carried out in 2 steps, wherein the first step, the treatment and outcome are modeled, and in the second step the predictions are inserted into the ATE estimator. In the first steps, numerous models can be fit to the treatment and outcome, including using machine learning algorithms. However, it is a difficult task to choose among the hyperparameter sets which will result in the best causal effect estimation and inference. Multiply Robust (MR) estimator allows us to leverage all the first-step models in a single estimator. We show that MR estimator is $n^r$ consistent if one of the first-step treatment or outcome models is $n^r$ consistent. We also show that MR is the solution to a broad class of estimating equations, and is asymptotically normal if one of the treatment models is $\sqrt{n}$-consistent. The standard error of MR is also calculated which does not require a knowledge of the true models in the first step. Our simulations study supports the theoretical findings.

We provide an analytical characterization of the coverage region of simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR-RIS)-aided two-user downlink communication systems. The cases of orthogonal multiple access (OMA) and non-orthogonal multiple access (NOMA) are considered, under the energy-splitting (ES) protocol. Results confirm that the use of STAR-RISs is beneficial to extend the coverage region, and that the use of NOMA provides a better performance compared to the OMA counterpart.