arXiv subjects

Ayush Prasad

Publications and source records attributed to Ayush Prasad.

4 recordsLinked to original sources

arXiv2026

Moonstone: A Multimodal Foundation Model and Benchmark for Lunar Remote Sensing

Decades of orbital missions have produced multi-modal remote sensing data for the Moon, spanning optical imagery, spectroscopy, thermal emission, radar, gravity, and elemental composition. Yet these datasets remain fragmented across archives, and no benchmark exists for evaluating machine learning on lunar data. We introduce Moonstone, the first multi-modal foundation model benchmark for lunar remote sensing. Our contributions are: (1) a 28-channel, 128 pixels-per-degree (~237 m) global lunar pretraining dataset from seven instrument families across five missions, (2) MG-MAE, a modality-grouped masked autoencoder with per-group convolutional tokenizers, a shared Vision Transformer encoder, attention masking for missing modalities, coverage-adaptive masking for heterogeneous spatial coverage, and spectral continuity regularization for physically plausible reconstructions, and (3) a benchmark of six downstream tasks covering classification, regression, and segmentation. MG-MAE pretrained features outperform scratch baselines on all tasks and surpass both ImageNet-pretrained and vanilla MAE baselines by large margins. Data and code are available at https://huggingface.co/datasets/ayushprd/Moonstone and https://github.com/ayushprd/Moonstone .

cs.CV↗

arXiv2024

Modeling Snow on Sea Ice using Physics Guided Machine Learning

Snow is a crucial element of the sea ice system, affecting sea ice growth and decay due to its low thermal conductivity and high albedo. Despite its importance, present-day climate models have an idealized representation of snow, often including only single-layer thermodynamics and omitting several processes that shape its properties. Although advanced snow process models like SnowModel exist, they are often excluded from climate modeling due to their high computational costs. SnowModel simulates snow depth, density, blowing-snow redistribution, sublimation, grain size, and thermal conductivity in a multi-layer snowpack. It operates with high spatial (1 meter) and temporal (1 hour) resolution. However, for large regions like the Arctic Ocean, these high-resolution simulations face challenges such as slow processing and large resource requirements. Data-driven emulators are used to address these issues, but they often lack generalizability and consistency with physical laws. In our study, we address these challenges by developing a physics-guided emulator that incorporates physical laws governing changes in snow density due to compaction. We evaluated three machine learning models: Long Short-Term Memory (LSTM), Physics-Guided LSTM, and Random Forest across five Arctic regions. All models achieved high accuracy, with the Physics-Guided LSTM showing the best performance in accuracy and generalizability. Our approach offers a faster way to emulate SnowModel with a speedup of over 9000 times, maintaining high fidelity.

physics.ao-ph↗

arXiv2024

Evaluating the transferability potential of deep learning models for climate downscaling

Climate downscaling, the process of generating high-resolution climate data from low-resolution simulations, is essential for understanding and adapting to climate change at regional and local scales. Deep learning approaches have proven useful in tackling this problem. However, existing studies usually focus on training models for one specific task, location and variable, which are therefore limited in their generalizability and transferability. In this paper, we evaluate the efficacy of training deep learning downscaling models on multiple diverse climate datasets to learn more robust and transferable representations. We evaluate the effectiveness of architectures zero-shot transferability using CNNs, Fourier Neural Operators (FNOs), and vision Transformers (ViTs). We assess the spatial, variable, and product transferability of downscaling models experimentally, to understand the generalizability of these different architecture types.

cs.LG↗

arXiv2024

Gradient Boosting Mapping for Dimensionality Reduction and Feature Extraction

A fundamental problem in supervised learning is to find a good set of features or distance measures. If the new set of features is of lower dimensionality and can be obtained by a simple transformation of the original data, they can make the model understandable, reduce overfitting, and even help to detect distribution drift. We propose a supervised dimensionality reduction method Gradient Boosting Mapping (GBMAP), where the outputs of weak learners -- defined as one-layer perceptrons -- define the embedding. We show that the embedding coordinates provide better features for the supervised learning task, making simple linear models competitive with the state-of-the-art regressors and classifiers. We also use the embedding to find a principled distance measure between points. The features and distance measures automatically ignore directions irrelevant to the supervised learning task. We also show that we can reliably detect out-of-distribution data points with potentially large regression or classification errors. GBMAP is fast and works in seconds for dataset of million data points or hundreds of features. As a bonus, GBMAP provides a regression and classification performance comparable to the state-of-the-art supervised learning methods.

cs.LG↗