Search

This technical report presents Prithvi-EO-2.0, a new geospatial foundation model that offers significant improvements over its predecessor, Prithvi-EO-1.0. Trained on 4.2M global time series samples from NASA's Harmonized Landsat and Sentinel-2 data archive at 30m resolution, the new 300M and 600M parameter models incorporate temporal and location embeddings for enhanced performance across various geospatial tasks. Through extensive benchmarking with GEO-Bench, the 600M version outperforms the previous Prithvi-EO model by 8\% across a range of tasks. It also outperforms six other geospatial foundation models when benchmarked on remote sensing tasks from different domains and resolutions (i.e. from 0.1m to 15m). The results demonstrate the versatility of the model in both classical earth observation and high-resolution applications. Early involvement of end-users and subject matter experts (SMEs) are among the key factors that contributed to the project's success. In particular, SME involvement allowed for constant feedback on model and dataset design, as well as successful customization for diverse SME-led applications in disaster response, land use and crop mapping, and ecosystem dynamics monitoring. Prithvi-EO-2.0 is available on Hugging Face and IBM terratorch, with additional resources on GitHub. The project exemplifies the Trusted Open Science approach embraced by all involved organizations.

This paper examines the communication performance of unmanned aerial vehicles (UAVs) in dense urban environments, specifically in Benidorm, Spain. Through a comprehensive measurement campaign, we assessed key performance indicators (KPIs) relating to received signal strength and quality as well as rate across various locations, altitudes, operators, technologies, and frequencies, using different measurement equipment. The results highlight significant challenges, primarily due to the lack of planning for aerial coverage and interference, revealing that current cellular networks may fall short in supporting U-space communication needs. The paper calls for network upgrades to ensure reliable UAV operations in urban airspace, contributing to the integration of UAVS in urban logistics and mobility.

Deep learning-based methods have been widely researched in the areas of language and vision, demonstrating their capacity to understand long sequences of data and their usefulness in numerous helio-physics applications. Foundation models (FMs), which are pre-trained on a large-scale datasets, form the basis for a variety of downstream tasks. These models, especially those based on transformers in vision and language, show exceptional potential for adapting to a wide range of downstream applications. In this paper, we provide our perspective on the criteria for designing an FM for heliophysics and associated challenges and applications using the Solar Dynamics Observatory (SDO) dataset. We believe that this is the first study to design an FM in the domain of heliophysics., Comment: 31 Pages, 12 figures

Triggered by the realization that AI emulators can rival the performance of traditional numerical weather prediction models running on HPC systems, there is now an increasing number of large AI models that address use cases such as forecasting, downscaling, or nowcasting. While the parallel developments in the AI literature focus on foundation models -- models that can be effectively tuned to address multiple, different use cases -- the developments on the weather and climate side largely focus on single-use cases with particular emphasis on mid-range forecasting. We close this gap by introducing Prithvi WxC, a 2.3 billion parameter foundation model developed using 160 variables from the Modern-Era Retrospective Analysis for Research and Applications, Version 2 (MERRA-2). Prithvi WxC employs an encoder-decoder-based architecture, incorporating concepts from various recent transformer models to effectively capture both regional and global dependencies in the input data. The model has been designed to accommodate large token counts to model weather phenomena in different topologies at fine resolutions. Furthermore, it is trained with a mixed objective that combines the paradigms of masked reconstruction with forecasting. We test the model on a set of challenging downstream tasks namely: Autoregressive rollout forecasting, Downscaling, Gravity wave flux parameterization, and Extreme events estimation. The pretrained model with 2.3 billion parameters, along with the associated fine-tuning workflows, has been publicly released as an open-source contribution via Hugging Face.

In this article, we introduce a method to optimize 5G massive multiple-input multiple-output (mMIMO) connectivity for unmanned aerial vehicles (UAVs) on aerial highways through strategic cell association. UAVs operating in 3D space encounter distinct channel conditions compared to traditional ground user equipment (gUE); under the typical line of sight (LoS) condition, UAVs perceive strong reference signal received power (RSRP) from multiple cells within the network, resulting in a large set of suitable serving cell candidates and in low signal-to-interference-plus-noise ratio (SINR) due to high interference levels. Additionally, a downside of aerial highways is to pack possibly many UAVs along a small portion of space which, when taking into account typical LoS propagation conditions, results in high channel correlation and severely limits spatial multiplexing capabilities. In this paper, we propose a solution to both problems based on the suitable selection of serving cells based on a new metric which differs from the classical terrestrial approaches based on maximum RSRP. We then introduce an algorithm for optimal planning of synchronization signal block (SSB) beams for this set of cells, ensuring maximum coverage and effective management of UAVs cell associations. Simulation results demonstrate that our approach significantly improves the rates of UAVs on aerial highways, up to four times in achievable data rates, without impacting ground user performance.

Foundation models (FMs) currently dominate news headlines. They employ advanced deep learning architectures to extract structural information autonomously from vast datasets through self-supervision. The resulting rich representations of complex systems and dynamics can be applied to many downstream applications. Therefore, FMs can find uses in electric power grids, challenged by the energy transition and climate change. In this paper, we call for the development of, and state why we believe in, the potential of FMs for electric grids. We highlight their strengths and weaknesses amidst the challenges of a changing grid. We argue that an FM learning from diverse grid data and topologies could unlock transformative capabilities, pioneering a new approach in leveraging AI to redefine how we manage complexity and uncertainty in the electric grid. Finally, we discuss a power grid FM concept, namely GridFM, based on graph neural networks and show how different downstream tasks benefit., Comment: Major equal contributors: H.F.H., T.B., B.G., L.S.A.M., A.P., A.V., J.W.; Significant equal contributors: J.B., A.B.M., S.C., I.F., B.H., R.J., K.K., V.M., F.M., M.D.M., O.R., H.S., L.X., E.S.Y., A.Z.; Other equal contributors: A.J.B., R.J.B., B.P.B., J.S., S.S; Lead contact: H.F.H

The roadmap is organized into several thematic sections, outlining current computing challenges, discussing the neuromorphic computing approach, analyzing mature and currently utilized technologies, providing an overview of emerging technologies, addressing material challenges, exploring novel computing concepts, and finally examining the maturity level of emerging technologies while determining the next essential steps for their advancement., Comment: 90 pages, 22 figures, roadmap, neuromorphic

Methods.We applied distinct selection criteria to attain an SFG sample with minimal AGN contamination. Multiple approaches were used to estimate the intrinsic extinction, SFR and gas-phase metallicity for the SFGs. In conjunction with findings in the literature, we examined the correlation between SFRs and stellar masses ($M_*$), as well as the metallicity evolution depending on $M_*$. Finally, the 3D relationship between $M_*$, SFR and metallicity, is also studied.} Results. From the initial spectroscopic sample of 409 FIR-selected objects from the Lockman--SpReSO catalogue, 69 (17\%) AGNs have been identified and excluded, which is nearly double the percentage found in local studies, leaving a sample of 340 SFGs. The analysis of the $M_*$--SFR relationship revealed that Lockman--SpReSO IR-selected SFGs show signs of evolution at redshifts $z>0.4$, shifting above the main sequence, with a mean value of $\sim0.4$ dex. They are located within the starburst galaxy region since 78\% of the galaxies fall into this category. In addition, no evident flattening was found in the relation to specific SFR with redshift for $\log M_* (M_\odot) \gtrsim 10.5$. In line with the $M_*$--metallicity relation (MZR) outcomes published in previous studies for optically selected SFGs, however, during the analysis of the MZR, it was found that IR-selected SFGs exhibit lower metallicities than those anticipated on the basis of their $M_*$ and redshift. During the investigation of the 3D $M_*$--SFR--metallicity relation (FP), it was established that the research sample is consistent with relations in the existing literature, with an average scatter of $\sim0.2$ dex. However, a re-calibration of the FP when using the SFR obtained from the IR luminosity is required and, in this case, no attenuation in the correlation for $\log M_* (M_\odot) \gtrsim 10.5$ is observed., Comment: 21 pages, 17 figures, accepted in A&A

Configurable synaptic delays are a basic feature in many neuromorphic neural network hardware accelerators. However, they have been rarely used in model implementations, despite their promising impact on performance and efficiency in tasks that exhibit complex (temporal) dynamics, as it has been unclear how to optimize them. In this work, we propose a framework to train and deploy, in digital neuromorphic hardware, highly performing spiking neural network models (SNNs) where apart from the synaptic weights, the per-synapse delays are also co-optimized. Leveraging spike-based back-propagation-through-time, the training accounts for both platform constraints, such as synaptic weight precision and the total number of parameters per core, as a function of the network size. In addition, a delay pruning technique is used to reduce memory footprint with a low cost in performance. We evaluate trained models in two neuromorphic digital hardware platforms: Intel Loihi and Imec Seneca. Loihi offers synaptic delay support using the so-called Ring-Buffer hardware structure. Seneca does not provide native hardware support for synaptic delays. A second contribution of this paper is therefore a novel area- and memory-efficient hardware structure for acceleration of synaptic delays, which we have integrated in Seneca. The evaluated benchmark involves several models for solving the SHD (Spiking Heidelberg Digits) classification task, where minimal accuracy degradation during the transition from software to hardware is demonstrated. To our knowledge, this is the first work showcasing how to train and deploy hardware-aware models parameterized with synaptic delays, on multicore neuromorphic hardware accelerators.

As nascent nonlinear oscillators, nano-constriction spin Hall nano-oscillators (SHNOs) represent a promising potential for integration into more complicated systems such as neural networks, magnetic field sensors, and radio frequency (RF) signal classification, their tunable high-frequency operating regime, easy synchronization, and CMOS compatibility can streamline the process. To implement SHNOs in any of these networks, the electrical features of a single device are needed before designing the signal detection CMOS circuitry. This study centers on presenting an empirical electrical model of the SHNO based on a comprehensive characterization of the output impedance of a single SHNO, and its available output power in the range of 2-10 GHz at various bias currents., Comment: 6 pages

In this paper, we revisit the data for the galaxy cluster ZwCl 0024.0+1652 provided by the GLACE survey and study the mass--metallicity function and its relationship with the environment. Here we describe an alternative way to reduce the data from OSIRIS tunable filters. This method gives us better uncertainties in the fluxes of the emission lines and the derived quantities. We present an updated catalogue of cluster galaxies with emission in H$\alpha$ and [N\,{\sc{ii}}] $\lambda\lambda$6548,6583. We also discuss the biases of these new fluxes and describe the way in which we calculated the mass--metallicity relationship and its uncertainties. We generated a new catalogue of 84 emission-line galaxies with reliable fluxes in [N\,{\sc{ii}}] and H$\alpha$ lines from a list of 174 galaxies. We find a relationship between the clustercentric radius and the density of galaxies. We derived the mass--metallicity relationship for ZwCl 0024.0+1652 and compared it with clusters and field galaxies from the literature. We find a difference in the mass--metallicity relationship when compared to more massive clusters, with the latter showing on average higher values of abundance. This could be an effect of the quenching of the star formation, which seems to be more prevalent in low-mass galaxies in more massive clusters. We find little to no difference between ZwCl 0024.0+1652 galaxies and field galaxies located at the same redshift., Comment: 13 pages, 17 figures, accepted for publication in Astronomy and Astrophysics

Epilepsy is a prevalent neurological disorder that affects approximately 1% of the global population. Around 30-40% of patients do not respond to pharmacological treatment, leading to a significant negative impact on their quality of life. Closed-loop deep brain stimulation (DBS) is a promising treatment for individuals who do not respond to medical therapy. To achieve effective seizure control, algorithms play an important role in identifying relevant electrographic biomarkers from local field potentials (LFPs) to determine the optimal stimulation timing. In this regard, the detection and classification of events from ongoing brain activity, while achieving low power through computationally unexpensive implementations, represents a major challenge in the field. To address this challenge, we here present two lightweight algorithms, the ZdensityRODE and the AMPDE, for identifying relevant events from LFPs by utilizing semantic segmentation, which involves extracting different levels of information from the LFP and relevant events from it. The algorithms performance was validated against epileptiform activity induced by 4-minopyridine in mouse hippocampus-cortex (CTX) slices and recorded via microelectrode array, as a case study. The ZdensityRODE algorithm showcased a precision and recall of 93% for ictal event detection and 42% precision for interictal event detection, while the AMPDE algorithm attained a precision of 96% and recall of 90% for ictal event detection and 54% precision for interictal event detection. While initially trained specifically for detection of ictal activity, these algorithms can be fine-tuned for improved interictal detection, aiming at seizure prediction. Our results suggest that these algorithms can effectively capture epileptiform activity; their light weight opens new possibilities for real-time seizure detection and seizure prediction and control., Comment: submitted to Plos One

Samples of $\alpha$-SiC have been sintered with liquid phase (LPS) to 1950 $^\circ$C, in atmosphere of argon and processing times between 1 and 7 hours. Using scanning electron microscopy (SEM) the microstructure of the samples has been characterized, being obtained average grain size that grows with the processing time from 0.64 to 1.61 $\mu$m. These samples have been deformed in compression under constant load, to temperatures between 1450 and 1625 C, stresses between 25 and 450 MPa and strain rate between $4.2\times 10^{-8}$ and $1.5\times 10^{-6}$ s$^{-1}$. The creep parameters have been determined, obtaining for n values between 2.4 0.1 and 4.5 0.2 and Q=680 35 kJ.mol$^{-1}$ in samples sintered during 1 hour and n between 1.2 0.1 and 2.4 0.1 and Q=710 90 kJ.mol$^{-1}$ for samples sintered during 7 hours. These results have been correlated with the microstructure, being the grain boundary sliding (GBS) accommodated by lattice diffusion and the movement of dislocations the possible deformation mechanisms, operating both in a simultaneous way., Comment: in Spanish language (Microestructura y propiedades mec\'anicas a altas temperaturas del $\alpha$-SiC sinterizado con fase l\'iquida de Y$_2$O$_3$-Al$_2$O$_3$)

The plastic deformation of Cr3+ -doped a-Al2 O3 (ruby) with four distinct Cr concentrations has been studied at temperatures between 900 and 1500 $^\circ$C. The mechanical tests indicate that the processes of dislocation multiplication and the adjustments of the slip velocity of these dislocations to the imposed strain rate in undoped a-Al2 O3 are not significantly affected by the Cr concentra%on. The yield stress increment is approximately constant with temperature and the hardening increases with the Cr concentration. The critical resolved shear stress (CRSS) is fixed to a model based on dislocation glide controlled by the nucleation and propagation of kink pairs, modified to include the hardening due to Cr. Satisfactory fits are achieved with adjustable parameters close to those employed for undoped a-Al2 O3 .

The influence that the atmosphere (N_2 or Ar) and sintering time have on microstructure evolution in liquid-phase-sintered alpha-sic and on its mechanical properties at high temperature was investigated. The microstructure of the samples sintered in N2 was equiaxed with a grain size of 0.70 {\mu}m and a density of 98% of the theoretical value regardless of the sintering time. In contrast, samples sintered in Ar had an elongated-grain microstructure with a density decreasing from 99% to 95% and a grain size increasing from 0.64 to 1.61 {\mu}m as the sintering time increased from 1 to 7 hours. The mechanical behaviour at 1450 {\deg}C showed the samples sintered in nitrogen to be brittle and fail at very low strains, with a fracture stress increasing from 400 to 800 MPa as the sintering time increased. In contrast, the samples sintered in Ar were quasi-ductile with increasing strain to failure as the sintering time increased, and a fracture stress strongly linked to the form and size of the grains. These differences in the mechanical properties of the two materials are discussed in the text. During mechanical tests, a loss of intergranular phase takes place in a region, between 50 and 150 {\mu}m thick, close to the surface of the samples--the effect being more important in the samples sintered in Ar, Comment: 27 pages, 8 figures

A model based on dislocation glide controlled by the nucleation and propagation of kink pairs in a high Peierls stress crystal is revisited and modified to account for changes in dislocation densities and segment lengths with temperature and stress. It is applied to the critical resolved shear stress (CRSS) for basal and prism plane slip in sapphire (alpha-Al2O3). According to agreed-upon knowledge on dislocations in sapphire, basal slip and prism plane slip are modelled with undissociated and dissociated dislocations, respectively. In the latter case, partial dislocations move independently. Amongst a number of sets of fitting parameters, good fits between experimental and modelled CRSS9s are obtained in the long segment limit over the whole range of temperatures by making use of physically sound parameters, including a stress dependence of the dislocation density.

Methods. We performed measurements of the \MgII, \MgI, \FeIIa, \FeIIb, and \FeIIc\ spectral lines present in the spectra of the selected sample to determine the EW and velocity of the flows observed in the star-forming galaxies. Subsequently, we conducted $10^7$ bootstrap simulations using Spearman's rank correlation coefficient ($\rho_s$) to explore correlations with galaxy properties. Furthermore, we calculated the covering factor, gas density, and optical depth for the measured \ion{Fe}{II} doublets. Results. Our analysis revealed strong correlations between the EW of \ion{Mg}{II} lines and both $M_{*}$ ($\rho_s=0.43$, 4.5$\sigma$) and SFR ($\rho_s=0.42$, 4.4$\sigma$). For the \ion{Fe}{II} lines, we observed strong correlations between the EW and SFR ($\rho_s\sim0.65$, $>3.9\sigma$), with a weaker correlation for $M_{*}$ ($\rho_s\sim0.35$, $>1.9\sigma$). No notable correlations were found between velocity measurements of \ion{Mg}{II} line and $M_{*}$, SFR, or sSFR of the objects ($\rho_s\sim0.1)$. However, a negative strong correlation was found between the velocity of the \ion{Fe}{II} lines and the SFR of the galaxies ($\rho_s\sim-0.45$, $\sim3\sigma$). Our results align with previous studies but studying FIR-selected objects. Finally, we detected a candidate \textit{loitering outflow}, a recently discovered subtype of FeLoBAL quasar, at redshift of $z=1.4399$, exhibiting emission in \ion{C}{III}] and low line velocities ($|v|\lesssim$ 200 km/s)., Comment: 25 pages, 31 figures

Hyper-trabeculation or non-compaction in the left ventricle of the myocardium (LVNC) is a recently classified form of cardiomyopathy. Several methods have been proposed to quantify the trabeculae accurately in the left ventricle, but there is no general agreement in the medical community to use a particular approach. In previous work, we proposed DL-LVTQ, a deep learning approach for left ventricular trabecular quantification based on a U-Net CNN architecture. DL-LVTQ was an automatic diagnosis tool developed from a dataset of patients with the same cardiomyopathy (hypertrophic cardiomyopathy). In this work, we have extended and adapted DL-LVTQ to cope with patients with different cardiomyopathies. The dataset consists of up 379 patients in three groups with different particularities and cardiomyopathies. Patient images were taken from different scanners and hospitals. We have modified and adapted the U-Net convolutional neural network to account for the different particularities of a heterogeneous group of patients with various unclassifiable or mixed and inherited cardiomyopathies. The inclusion of new groups of patients has increased the accuracy, specificity and kappa values while maintaining the sensitivity of the automatic deep learning method proposed. Therefore, a better-prepared diagnosis tool is ready for various cardiomyopathies with different characteristics. Cardiologists have considered that 98.9% of the evaluated outputs are verified clinically for diagnosis. Therefore, the high precision to segment the different cardiac structures allows us to make a robust diagnostic system objective and faster, decreasing human error and time spent.

This paper presents the real-world smart-meter dataset and offers an analysis of solutions derived from the Energy Prediction Technical Challenges, focusing primarily on two key competitions: the IEEE Computational Intelligence Society (IEEE-CIS) Technical Challenge on Energy Prediction from Smart Meter data in 2020 (named EP) and its follow-up challenge at the IEEE International Conference on Fuzzy Systems (FUZZ-IEEE) in 2021 (named as XEP). These competitions focus on accurate energy consumption forecasting and the importance of interpretability in understanding the underlying factors. The challenge aims to predict monthly and yearly estimated consumption for households, addressing the accurate billing problem with limited historical smart meter data. The dataset comprises 3,248 smart meters, with varying data availability ranging from a minimum of one month to a year. This paper delves into the challenges, solutions and analysing issues related to the provided real-world smart meter data, developing accurate predictions at the household level, and introducing evaluation criteria for assessing interpretability. Additionally, this paper discusses aspects beyond the competitions: opportunities for energy disaggregation and pattern detection applications at the household level, significance of communicating energy-driven factors for optimised billing, and emphasising the importance of responsible AI and data privacy considerations. These aspects provide insights into the broader implications and potential advancements in energy consumption prediction. Overall, these competitions provide a dataset for residential energy research and serve as a catalyst for exploring accurate forecasting, enhancing interpretability, and driving progress towards the discussion of various aspects such as energy disaggregation, demand response programs or behavioural interventions.

In maritime traffic surveillance, detecting illegal activities, such as illegal fishing or transshipment of illicit products is a crucial task of the coastal administration. In the open sea, one has to rely on Automatic Identification System (AIS) message transmitted by on-board transponders, which are captured by surveillance satellites. However, insincere vessels often intentionally shut down their AIS transponders to hide illegal activities. In the open sea, it is very challenging to differentiate intentional AIS shutdowns from missing reception due to protocol limitations, bad weather conditions or restricting satellite positions. This paper presents a novel approach for the detection of abnormal AIS missing reception based on self-supervised deep learning techniques and transformer models. Using historical data, the trained model predicts if a message should be received in the upcoming minute or not. Afterwards, the model reports on detected anomalies by comparing the prediction with what actually happens. Our method can process AIS messages in real-time, in particular, more than 500 Millions AIS messages per month, corresponding to the trajectories of more than 60 000 ships. The method is evaluated on 1-year of real-world data coming from four Norwegian surveillance satellites. Using related research results, we validated our method by rediscovering already detected intentional AIS shutdowns., Comment: IEEE Transactions on Intelligent Transportation Systems

The role of axonal synaptic delays in the efficacy and performance of artificial neural networks has been largely unexplored. In step-based analog-valued neural network models (ANNs), the concept is almost absent. In their spiking neuroscience-inspired counterparts, there is hardly a systematic account of their effects on model performance in terms of accuracy and number of synaptic operations.This paper proposes a methodology for accounting for axonal delays in the training loop of deep Spiking Neural Networks (SNNs), intending to efficiently solve machine learning tasks on data with rich temporal dependencies. We then conduct an empirical study of the effects of axonal delays on model performance during inference for the Adding task, a benchmark for sequential regression, and for the Spiking Heidelberg Digits dataset (SHD), commonly used for evaluating event-driven models. Quantitative results on the SHD show that SNNs incorporating axonal delays instead of explicit recurrent synapses achieve state-of-the-art, over 90% test accuracy while needing less than half trainable synapses. Additionally, we estimate the required memory in terms of total parameters and energy consumption of accomodating such delay-trained models on a modern neuromorphic accelerator. These estimations are based on the number of synaptic operations and the reference GF-22nm FDX CMOS technology. As a result, we demonstrate that a reduced parameterization, which incorporates axonal delays, leads to approximately 90% energy and memory reduction in digital hardware implementations for a similar performance in the aforementioned task.

Large language models (LLMs) have been massively applied to many tasks, often surpassing state-of-the-art approaches. While their effectiveness in code generation has been extensively studied (e.g., AlphaCode), their potential for code detection remains unexplored. This work presents the first analysis of code detection using LLMs. Our study examines essential kernels, including matrix multiplication, convolution, and fast-fourier transform, implemented in C/C++. We propose both a preliminary, naive prompt and a novel prompting strategy for code detection. Results reveal that conventional prompting achieves great precision but poor accuracy (68.8%, 22.3%, and 79.2% for GEMM, convolution, and FFT, respectively) due to a high number of false positives. Our novel prompting strategy substantially reduces false positives, resulting in excellent overall accuracy (91.1%, 97.9%, and 99.7%, respectively). These results pose a considerable challenge to existing state-of-the-art code detection methods.

In this paper we present a memristor-inspired computational method for obtaining a type of running spectrogram or fingerprint of epileptiform activity generated by rodent hippocampal spheroids. It can be used to compute on the fly and with low computational cost an alert-level signal for epileptiform events onset. Here, we describe the computational method behind this fingerprint technique and illustrate it using epileptiform events recorded from hippocampal spheroids using a microelectrode array system., Comment: published in 2023 IEEE 5th International Conference on Artificial Intelligence Circuits and Systems (AICAS)