Search

Machine learning models often experience performance degradation post-deployment due to shifts in data distribution. It is challenging to assess model's performance accurately when labels are missing or delayed. Existing proxy methods, such as drift detection, fail to measure the effects of these shifts adequately. To address this, we introduce a new method, Probabilistic Adaptive Performance Estimation (PAPE), for evaluating classification models on unlabeled data that accurately quantifies the impact of covariate shift on model performance. It is model and data-type agnostic and works for various performance metrics. Crucially, PAPE operates independently of the original model, relying only on its predictions and probability estimates, and does not need any assumptions about the nature of the covariate shift, learning directly from data instead. We tested PAPE on tabular data using over 900 dataset-model combinations created from US census data, assessing its performance against multiple benchmarks. Overall, PAPE provided more accurate performance estimates than other evaluated methodologies., Comment: 9 content pages, 3 figures

Perovskite solar cells (PSCs) are the most promising technology for advancing current photovoltaic performance. However, the main challenge for their practical deployment and commercialization is their operational stability, affected by solar illumination and heating, as well as the electric field that is generated in the PV device by light exposure. Here, we propose a transparent multispectral photonic electrode placed on top of the glass substrate of solar cells, which simultaneously reduces the device solar heating and enhances its efficiency. Specifically, the proposed photonic electrode, composed of a low-resistivity metal and a conductive layer, simultaneously serves as a highly-efficient infrared filter and an ultra-thin transparent front contact, decreasing devices' solar heating and operating temperature. At the same time, it simultaneously serves as an anti-reflection coating, enhancing the efficiency. We additionally enhance the device cooling by coating the front glass substrate side with a visibly transparent film (PDMS), which maximizes substrate's thermal radiation. To determine the potential of our photonic approach and fully explore the cooling potential of PSCs, we first provide experimental characterizations of the absorption properties (in both visible and infrared wavelengths) of state-of-the-art PSCs among the most promising ones regarding the efficiency, stability, and cost. We then numerically show that applying our approach to promising PSCs can result in lower operating temperatures by over 9.0 oC and an absolute efficiency increase higher than 1.3%. These results are insensitive to varying environmental conditions. Our approach is simple and only requires modification of the substrate; it therefore points to a feasible photonic approach for advancing current photovoltaic performance with next-generation solar cell materials., Comment: 20 pages, 9 figures

Ultrathin, solution-processed emerging solar cells with high power-per-weight (PPW) outputs demonstrate unique potential for applications where low weight, high power output, and flexibility are indispensable. The following perspective explores the literature of emerging PVs and highlights the maximum reported PPW values of Perovskite Solar Cells (PSCs) 29.4 W/g, Organic Solar Cells (OSCs) 32.07 W/g and Quantum Dot Solar Cells (QDSC) 15.02 W/g, respectively. The record PPW values of OSCs and PSCs are approximately one order of magnitude higher compared to their inorganic ultrathin solar cells counterparts (approx. 3.2 W/g for CIGS and a-Si). This consists emerging PVs, very attractive for a variety of applications where the PPW is the key parameter. In particular, both OSCs and PSCs can be implemented in different scenarios of applications (indoor and biocompatible applications for OSCs and outdoor and high-energy radiation conversion conditions for the PSCs) due to their unique optoelectronic and physiochemical properties. Finally, our theoretical optical and electrical simulation and optimization study for the most promising and well-suited PV technologies, showed an impressive maximum realistic theoretical PPW limit of 74.3 and 93.7 W/g for PSCs and OSCs, respectively. Our finding shows that the literature PSCs and OSCs towards high PPW outputs, is not quite close to the theoretical maximum and thus more work needs to be done to further increase the PPW output of these promising PV technologies., Comment: 41 pages, 20 figues

Formation of nanoscale laser-induced periodic surface structures on thin metal films (of the size of the optical penetration depth) is a yet unexplored area that is expected to open new routes for laser patterning and a wealth of exciting applications in optics, photonics, and sensing. In contrast to the common belief that excitation of Surface Plasmon Polaritons (SPPs) on the air/metal interface plays the dominant role in the features of the induced topographies, in this work, we demonstrate that the excitation of coupled SPPs in both air/metal and metal/substrate interfaces, along with other parameters such as the thickness of the material, the photon energy, and the substrate refractive index, dictate the spatial modulation of the absorbed energy. A detailed theoretical analysis of the excited plasmonic waves and a multiscale modelling of laser-induced physical phenomena manifests that depending on the laser conditions and thickness of the irradiated solid, topographies with periodic features of diverse sizes (ranging from $\lambda_L/3$ to $\lambda_L$, where $\lambda_L$ stands for the laser wavelength) and different orientation can be realized. The capability to control and tune the characteristics of the produced structures on thin films is expected to enable novel surface patterning approaches., Comment: To be published in Optics and Laser Technology

Machine-learning based classifiers have become indispensable in the field of astrophysics, allowing separation of astronomical sources into various classes, with computational efficiency suitable for application to the enormous data volumes that wide-area surveys now typically produce. In the standard supervised classification paradigm, a model is typically trained and validated using data from relatively small areas of sky, before being used to classify sources in other areas of the sky. However, population shifts between the training examples and the sources to be classified can lead to `silent' degradation in model performance, which can be challenging to identify when the ground-truth is not available. In this Letter, we present a novel methodology using the NannyML Confidence-Based Performance Estimation (CBPE) method to predict classifier F1-score in the presence of population shifts, but without ground-truth labels. We apply CBPE to the selection of quasars with decision-tree ensemble models, using broad-band photometry, and show that the F1-scores are predicted remarkably well (MAPE ~ 10%; R^2 = 0.74-0.92). We discuss potential use-cases in the domain of astronomy, including machine-learning model and/or hyperparameter selection, and evaluation of the suitability of training datasets for a particular classification problem., Comment: 5.5 pages, 2 figures, accepted for publication in MNRAS Letters

In this paper we present Poisson mixture approaches for origin-destination (OD) modeling in transportation analysis. We introduce covariate-based models which incorporate different transport modeling phases and also allow for direct probabilistic inference on link traffic based on Bayesian predictions. Emphasis is placed on the Poisson-inverse Gaussian as an alternative to the commonly-used Poisson-gamma and Poisson-lognormal models. We present a first full Bayesian formulation and demonstrate that the Poisson-inverse Gaussian is particularly suited for OD analysis due to desirable marginal and hierarchical properties. In addition, the integrated nested Laplace approximation (INLA) is considered as an alternative to Markov chain Monte Carlo and the two methodologies are compared under specific modeling assumptions. The case study is based on 2001 Belgian census data and focuses on a large, sparsely-distributed OD matrix containing trip information for 308 Flemish municipalities.

We report on a single-step, biomimetic approach for the realization of omnidirectional transparent antireflective glass. In particular, it is shown that circularly polarized ultrashort laser pulses produce self-organized nanopillar structures on fused silica (SiO2). The laser induced nanostructures are selectively textured on the glass surface in order to mimic the spatial randomness, pillar-like morphology, as well as the remarkable anti-reflection properties found on the wings of the glasswing butterfly, Greta oto[1] and various Cicada species[2]. The artificial structures exhibit impressive anti-reflective properties, both in the visible and infrared frequency range. Accordingly, the laser-processed glass surfaces show reflectivity smaller than 1% for various angles of incidence in the visible spectrum for S-P linearly polarized configurations. While, in the near infrared spectrum, the laser-textured glass shows higher transmittance compared to the pristine. It is envisaged that our current results will revolutionize the technology of anti-reflective transparent surfaces and impact numerous applications from glass displays to optoelectronic devices.

We investigate and evaluate the contribution of the ultraviolet radiation spectrum on the temperature and efficiency of commercial crystalline silicon-based photovoltaics (PVs) that operate outdoors. The investigation is performed by employing a comprehensive thermal-electrical modeling approach which takes into account all the major processes affected by the temperature variation in the photovoltaic devices. We show that effectively reflecting the ultraviolet radiation (i.e. up to a certain wavelength) results in a reduction of the overall operation temperature and enhancement of the PV cell's efficiency. In addition, blocking the high energy ultraviolet photons prolongs the life time of the PV and its performance on the long term., Comment: 5 pages, 3 figures

The radiative cooling of objects during daytime under direct sunlight has recently been shown to be significantly enhanced by utilizing nanophotonic coatings. Multilayer thin film stacks, 2D photonic crystals, etc. as coating structures improved the thermal emission rate of a device in the infrared atmospheric transparency window reducing considerably devices' temperature. Due to the increased heating in photovoltaic (PV) devices, that has significant adverse consequences on both their efficiency and life-time, and inspired by the recent advances in daytime radiative cooling, we developed a coupled thermal-electrical modeling to examine the physical mechanisms on how a radiative cooler affects the overall efficiency of commercial photovoltaic modules. Employing this modeling, which takes into account all the major processes affected by the temperature variation in a PV device, we evaluated the relative impact of the main radiative cooling approaches proposed so far on the PV efficiency, and we established required conditions for optimized radiative cooling. Moreover, we identified the validity regimes of the currently existing PV-cooling models which treat the PV coolers as simple thermal emitters. Finally, we assessed some realistic photonic coolers from the literature, compatible with photovoltaics, to implement the radiative cooling requirements, and demonstrated their associated impact on the temperature reduction and PV efficiency. Providing the physical mechanisms and requirements for cooling radiatively solar cells, our study provides guidelines for utilizing suitable photonic structures as radiative coolers, enhancing the efficiency and the lifetime of PV devices., Comment: 20 pages, 5 figures

Regularized regression models are well studied and, under appropriate conditions, offer fast and statistically interpretable results. However, large data in many applications are heterogeneous in the sense of harboring distributional differences between latent groups. Then, the assumption that the conditional distribution of response Y given features X is the same for all samples may not hold. Furthermore, in scientific applications, the covariance structure of the features may contain important signals and its learning is also affected by latent group structure. We propose a class of mixture models for paired data (X, Y) that couples together the distribution of X (using sparse graphical models) and the conditional Y | X (using sparse regression models). The regression and graphical models are specific to the latent groups and model parameters are estimated jointly (hence the name "regularized joint mixtures"). This allows signals in either or both of the feature distribution and regression model to inform learning of latent structure and provides automatic control of confounding by such structure. Estimation is handled via an expectation-maximization algorithm, whose convergence is established theoretically. We illustrate the key ideas via empirical examples. An R package is available at https://github.com/k-perrakis/regjmix.

Solution-processed, lead halide-based perovskite solar cells have overcome important challenges over the recent years, offering low-cost and high solar power conversion efficiencies. However, they still undergo unoptimized light collection due mainly to the thin (~350 nm) polycrystalline absorber layers. Moreover, their high toxicity (due to the presence of lead in the perovskite crystalline structure) makes it necessary that the thickness of the absorber layers to be further reduced, for their future commercialization, without reducing the device performance. Here we aim to address these issues via embedding spherical plasmonic nanoparticles of various sizes, composition, concentrations, and vertical positions, for the first time in realistic halide-based perovskite solar cells architecture, and to clarify their effect on the absorption properties and enhancement. We theoretically show that plasmon-enhanced near-field effects and scattering leads to a device photocurrent enhancement of up to ~7.3% when silver spheres are embedded inside the perovskite layer. Interestingly, the combination of silver spheres in perovskite and aluminum spheres inside the hole transporting layer (PEDOT:PSS) of the solar cell leads to an even further enhancement, of up to ~12%. This approach allows the employment of much thinner perovskite layers in PSCs (up to 150 nm) to reach the same photocurrent as the nanoparticles-free device and reducing thus significantly the toxicity of the device. Providing the requirements related to the size, shape, position, composition, and concentration of nanoparticles for the PSCs photocurrent enhancement, our study establishes guidelines for a future development of highly-efficient, environmentally friendly and low-cost plasmonic perovskite solar cells., Comment: 16 pages, 9 figures (main text)

Applications of high-dimensional regression often involve multiple sources or types of covariates. We propose methodology for this setting, emphasizing the "wide data" regime with large total dimensionality p and sample size n<