Mgr. Petr Šimánek

Stať ve sborníku

10.15439/2024F5446

Katedra aplikované matematiky

Efficient crop monitoring and crop dynamics fore- casting leveraging diverse satellite and point data are described. Attention-based architecture architecture is adapted for mono- temporal cloud removal which overcomes an issue of crop monitoring. Combining optical (Sentinel-2) and radar (Sentinel- 1) satellite data improves the robustness and accuracy of the model in terms of satellite image reconstruction and normalized difference vegetation index prediction and forecasting. However, available soil-type geographical data and land surface analysis products, do not improve prediction accuracy significantly

Stať ve sborníku

10.15439/2024F5990

Katedra aplikované matematiky

Accurate local weather forecasting is vital for farmers to optimize crop yields and manage resources effectively, but existing forecasts often lack the precision required locally. This study explores the potential of combining data from local weather stations with global forecasts and reanalysis data to improve the accuracy of local weather predictions. We propose integrating the HadISD data set, which contains data from 27 stations in the Czech Republic, with the Global Forecast System predictions and ERA5-Land reanalysis data. Our goal is to improve 24-hour weather forecasts using Multilayer Perceptrons, CatBoost, and Long Short-Term Memory neural networks. The findings demonstrate that combining local weather station data with global forecasts and incorporating ERA5-Land reanalysis data improves the accuracy of weather predictions in specific locations. Notably, using deep learning to estimate ERA5-Land data and including this estimation in the final model reduced the forecasting error by 59\%. This advancement holds promise in optimizing agricultural practices and mitigating weather-related risks in the region.

Stať ve sborníku

10.1007/978-3-031-72356-8_13

Katedra aplikované matematiky
Fakulta informačních technologií

This study investigates the application of deep-learning diffusion models for the super-resolution of weather data, a novel approach aimed at enhancing the spatial resolution and detail of meteorological variables. Leveraging the capabilities of diffusion models, specifically the SR3 and ResDiff architectures, we present a methodology for transforming low-resolution weather data into high-resolution outputs. Our experiments, conducted using the WeatherBench dataset, focus on the super-resolution of the two-meter temperature variable, demonstrating the models’ ability to generate detailed and accurate weather maps. The results indicate that the ResDiff model, further improved by incorporating physics-based modifications, significantly outperforms traditional SR3 methods in terms of Mean Squared Error (MSE), Structural Similarity Index (SSIM), and Peak Signal-to-Noise Ratio (PSNR). This research highlights the potential of diffusion models in meteorological applications, offering insights into their effectiveness, challenges, and prospects for future advancements in weather prediction and climate analysis

Stať ve sborníku

10.5220/0012317000003636

Katedra aplikované matematiky
Fakulta informačních technologií

Optimization is an integral part of modern deep learning. Recently, the concept of learned optimizers has emerged as a way to accelerate this optimization process by replacing traditional, hand-crafted algorithms with meta-learned functions. Despite the initial promising results of these methods, issues with stability and generalization still remain, limiting their practical use. Moreover, their inner workings and behavior under different conditions are not yet fully understood, making it difficult to come up with improvements. For this reason, our work examines their optimization trajectories from the perspective of network architecture symmetries and parameter update distributions. Furthermore, by contrasting the learned optimizers with their manually designed counterparts, we identify several key insights that demonstrate how each approach can benefit from the strengths of the other.

Stať ve sborníku

10.1609/aaai.v38i21.30514

Katedra aplikované matematiky

Modern machine learning heavily relies on optimization, and as deep learning models grow more complex and data-hungry, the search for efficient learning becomes crucial. Learned optimizers disrupt traditional handcrafted methods such as SGD and Adam by learning the optimization strategy itself, potentially speeding up training. However, the learned optimizers' dynamics are still not well understood. To remedy this, our work explores their optimization trajectories from the perspective of network architecture symmetries and proposed parameter update distributions.

Článek

10.1007/s42979-024-02670-2

Katedra aplikované matematiky

The intuitive approach of comparing like with like, forms the basis of the so-called nearest neighbor analysis, which is central to many machine learning algorithms. Nearest neighbor analysis is easy to interpret, analyze, and reason about. It is widely used in advanced techniques such as uncertainty estimation in regression models, as well as the renowned k-nearest neighbor-based algorithms. Nevertheless, its high inference time complexity, which is dataset size dependent even in the case of its faster approximated version, restricts its applications and can considerably inflate the application cost. In this paper, we address the problem of high inference time complexity. By using gradient-boosted regression trees as a predictor of the labels obtained from nearest neighbor analysis, we demonstrate a significant increase in inference speed, improving by several orders of magnitude. We validate the effectiveness of our approach on a real-world European Car Pricing Dataset with approximately rows for both residual cost and price uncertainty prediction. Moreover, we assess our method’s performance on the most commonly used tabular benchmark datasets to demonstrate its scalability. The link is to github repository where the code is available: https://github.com/koutefra/uncertainty_experiments.

Stať ve sborníku

10.1162/isal_a_00831

Katedra aplikované matematiky

This study presents Dynamics Identification via Neural Cellular Automata (DINCA), an enhancement of Neural Cellular Automata (NCA) for modeling reaction-diffusion systems. The main advantage of DINCA is its ability to estimate the parameters of the reaction-diffusion equations that govern the examined system, using minimal data. We demonstrate the method’s application potential by showing its ability to model leopard pattern formation, by learning on only three images, while revealing the governing reaction-diffusion equations. This positions NCA-based methodologies as a viable tool for inferring partial differential equations.

Stať ve sborníku

Katedra aplikované matematiky

Weather4cast again advanced modern algorithms in AI and machine learning through a highly topical interdisciplinary competition challenge: The prediction of hi-res rain radar movies from multi-band satellite sensors, requiring data fusion, multi-channel video frame prediction, and super-resolution. Accurate predictions of rain events are becoming ever more critical, with climate change increasing the frequency of unexpected rainfall. The resulting models will have a particular impact where costly weather radar is not available. We here present highlights and insights emerging from the thirty teams participating from over a dozen countries. To extract relevant patterns, models were challenged by spatio-temporal shifts. Geometric data augmentation and test-time ensemble models with a suitable smoother loss helped this transfer learning. Even though, in ablation, static information like geographical location and elevation was not linked to performance, the general success of models incorporating physics in this competition suggests that approaches combining machine learning with application domain knowledge seem a promising avenue for future research. Weather4cast will continue to explore the powerful benchmark reference data set introduced here, advancing competition tasks to quantitative predictions, and exploring the effects of metric choice on model performance and qualitative prediction properties.

Stať ve sborníku

10.1109/IJCNN55064.2022.9892364

Katedra aplikované matematiky

Designing faster optimization algorithms is of ever-growing interest. In recent years, learning to learn methods that learn how to optimize demonstrated very encouraging results. Current approaches usually do not effectively include the dynamics of the optimization process during training. They either omit it entirely or only implicitly assume the dynamics of an isolated parameter. In this paper, we show how to utilize the dynamic mode decomposition method for extracting informative features about optimization dynamics. By employing those features, we show that our learned optimizer generalizes much better to unseen optimization problems in short. The improved generalization is illustrated on multiple tasks where training the optimizer on one neural network generalizes to different architectures and distinct datasets.

Stať ve sborníku

10.1109/IV51971.2022.9827089

Katedra aplikované matematiky
Fakulta informačních technologií

Prediction of the movement of all traffic participants is a very important task in autonomous driving. Well-predicted behavior of other cars and actors is crucial for safety. A sequence of bird’s-eye view artificially rasterized frames are used as input to neural networks which are trained to predict the future behavior of the participants. The Lyft Motion Prediction for Autonomous Vehicles dataset is explored and adapted for this task. We developed and applied a novel approach where the prediction problem is viewed as a problem of spatiotemporal prediction and we use methods based on convolutional recurrent neural networks.