检索结果-内蒙古大学图书馆

Sample Complexity of Kernel-Based Q-Learning

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Yeh, Sing-Yuan Chang, Fu-Chieh Yueh, Chang-Wei Wu, Pei-Yuan Bernacchia, Alberto Vakili, Sattar MediaTek Research Graduate Program of Data Science National Taiwan University Academia Sinica Taiwan Graduate Institute of Communication Engineering National Taiwan University Taiwan Department of Electrical Engineering National Taiwan University Taiwan

Modern reinforcement learning (RL) often faces an enormous state-action space. Existing analytical results are typically for settings with a small number of state-actions, or simple models such as linearly modeled Q-functions. To derive statistically efficient RL policies handling large state-action spaces, with more general Q-functions, some recent works have considered nonlinear function approximation using kernel ridge regression. In this work, we derive sample complexities for kernel based Q-learning when a generative model exists. We propose a nonparametric Q-learning algorithm which finds an ǫ-optimal policy in an arbitrarily large scale discounted MDP. The sample complexity of the proposed algorithm is order optimal with respect to ǫ and the complexity of the kernel (in terms of its information gain). To the best of our knowledge, this is the first result showing a finite sample complexity under such a general model. © 2023, CC BY.

关键词： Reinforcement learning

Graph Laplacian Learning with Exponential Family Noise

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Shi, Changhao Mishne, Gal The Electrical and Computer Engineering Department UC San Diego CA92093 United States The Halıcıoglu Data Science Institute The Neurosciences Graduate Program UC San Diego CA92093 United States

Graph signal processing (GSP) is a prominent framework for analyzing signals on non-Euclidean domains. The graph Fourier transform (GFT) uses the combinatorial graph Laplacian matrix to reveal the spectral decomposition of signals in the graph frequency domain. However, a common challenge in applying GSP methods is that in many scenarios the underlying graph of a system is unknown. A solution in such cases is to construct the unobserved graph from available data, which is commonly referred to as graph or network inference. Although different graph inference methods exist, these are restricted to learning from either smooth graph signals or simple additive Gaussian noise. Other types of noisy data, such as discrete counts or binary digits, are rather common in real-world applications, yet are underexplored in graph inference. In this paper, we propose a versatile graph inference framework for learning from graph signals corrupted by exponential family noise. Our framework generalizes previous methods from continuous smooth graph signals to various data types. We propose an alternating algorithm that jointly estimates the graph Laplacian and the unobserved smooth representation from the noisy signals. We also extend our approach to a variational form to account for the inherent stochasticity of the latent smooth representation. Finally, since real-world graph signals are frequently non-independent and temporally correlated, we further adapt our original setting to a time-vertex formulation. We demonstrate on synthetic and real-world data that our new algorithms outperform competing Laplacian estimation methods that suffer from noise model mismatch. Copyright © 2023, The Authors. All rights reserved.

关键词： Signal processing

Transfer Learning for Emulating Ocean Climate Variability across CO2 forcing

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Dheeshjith, Surya Subel, Adam Gupta, Shubham Adcroft, Alistair Fernandez-Granda, Carlos Busecke, Julius Zanna, Laure Courant Institute of Mathematical Sciences New York University United States Program in Atmospheric and Oceanic Sciences Princeton University United States Center for Data Science New York University United States Lamont Doherty Earth Observatory Columbia University United States

With the success of machine learning (ML) applied to climate reaching further every day, emulators have begun to show promise not only for weather but for multi-year time scales in the atmosphere. Similar work for the ocean remains nascent, with state-of-the-art limited to models running for shorter time scales or only for regions of the globe. In this work, we demonstrate high-skill global emulation for surface ocean fields over 5-8 years of model rollout, accurately representing modes of variability for two different ML architectures (ConvNext and Transformers). In addition, we address the outstanding question of generalization, an essential consideration if the end-use of emulation is to model warming scenarios outside of the model training data. We show that 1) generalization is not an intrinsic feature of a data-driven emulator, 2) fine-tuning the emulator on only small amounts of additional data from a distribution similar to the test set can enable the emulator to perform well in a warmed climate, and 3) the forced emulators are robust to noise in the forcing. © 2024, CC BY.

关键词： Transfer learning

Samudra: An AI Global Ocean Emulator for Climate

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Dheeshjith, Surya Subel, Adam Adcroft, Alistair Busecke, Julius Fernandez-Granda, Carlos Gupta, Shubham Zanna, Laure Courant Institute of Mathematical Sciences New York University United States Program in Atmospheric and Oceanic Sciences Princeton University United States Center for Data Science New York University United States Lamont Doherty Earth Observatory Columbia University United States

AI emulators for forecasting have emerged as powerful tools that can outperform conventional numerical predictions. The next frontier is to build emulators for long climate simulations with skill across a range of spatiotemporal scales, a particularly important goal for the ocean. Our work builds a skillful global emulator of the ocean component of a state-of-the-art climate model. We emulate key ocean variables, sea surface height, horizontal velocities, temperature, and salinity, across their full depth. We use a modified ConvNeXt UNet architecture trained on multi-depth levels of ocean data. We show that the ocean emulator – Samudra – which exhibits no drift relative to the truth, can reproduce the depth structure of ocean variables and their interannual variability. Samudra is stable for centuries and 150 times faster than the original ocean model. Samudra struggles to capture the correct magnitude of the forcing trends and simultaneously remain stable, requiring further work. © 2024, CC BY.

关键词： Oceanography

Classifying different types of solar wind plasma with uncertainty estimations using machine learning

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Narock, Tom Pal, Sanchita Arsham, Aryana Narock, Ayris Nieves-Chinchilla, Teresa Department of Computer and Data Science Goucher College BaltimoreMD21204 United States NASA Postdoctoral Program Fellow NASA Goddard Space Flight Center GreenbeltMD20771 United States ADNET Systems Inc BethesdaMD20817 United States Heliophysics Science Division NASA Goddard Space Flight Center GreenbeltMD20771 United States

Decades of in-situ solar wind measurements have clearly established the variation of solar wind physical parameters. These variable parameters have been used to classify the solar wind magnetized plasma into different types leading to several classification schemes being developed. These classification schemes, while useful for understanding the solar wind’s originating processes at the Sun and early detection of space weather events, have left open questions regarding which physical parameters are most useful for classification and how recent advances in our understanding of solar wind transients impact classification. In this work, we use neural networks trained with different solar wind magnetic and plasma characteristics to automatically classify the solar wind in coronal hole, streamer belt, sector reversal and solar transients such as coronal mass ejections comprised of both magnetic obstacles and sheaths. Furthermore, our work demonstrates how probabilistic neural networks can enhance the classification by including a measure of prediction uncertainty. Our work also provides a ranking of the parameters that lead to an improved classification scheme with ∼ 96% accuracy. Our new scheme paves the way for incorporating uncertainty estimates into space weather forecasting with the potential to be implemented on real-time solar wind data. © 2024, CC BY.

关键词： Solar wind

Meta-Learning for Generalizable Connectome Modeling Across Heterogeneous Atlas Spaces

学校读者我要写书评

暂无评论

Meta-Learning for Generalizable Connectome Modeling Across H...

IEEE International Symposium on Biomedical Imaging

作者： Qinghao Liang Brendan D. Adkinson Dustin Scheinost Department of Biomedical Engineering Yale University New Haven CT Interdepartmental Neuroscience Program Yale University New Haven CT Department of Radiology and Biomedical Imaging Yale School of Medicine New Haven CT Department of Statistics and Data Science Yale University New Haven CT

ISBN: (数字)9798331520526

ISBN: (纸本)9798331520533

Leveraging large-scale neuroimaging datasets to improve predictive modeling in smaller, specialized cohorts is essential for translating research findings into clinical applications. To address this challenge, we propose a meta-learning inspired approach that enhances phenotype prediction by exploiting inter-phenotype correlations across independent datasets, without requiring uniform imaging data processing. By employing mappings derived from a small, independent cohort with data processed using different atlases, we align the data in the meta-training set to the target space of the meta-testing set. This approach eliminates the need for extensive data reprocessing. Evaluation in the UK Biobank dataset shows that our framework significantly improves predictive performance on unseen phenotypes, demonstrating its potential to generalize connectome-based models to small, heterogeneous cohorts.

关键词： Metalearning Neuroimaging Phenotypes Correlation Biological system modeling Predictive models data processing Biomedical imaging

Early-Stage Detection of Cognitive Impairment by Hybrid Quantum-Classical Algorithm Using Resting-State Functional MRI Time-Series

学校读者我要写书评

暂无评论

SSRN

SSRN 2024年

作者： Choi, Junggu Hur, Tak Park, Daniel Kyungdeock Shin, Na-Young Lee, Seung-Koo Lee, Hakbae Han, Sanghoon Yonsei Graduate Program in Cognitive Science Yonsei University Seoul Korea Republic of Department of Statistics and Data Science Yonsei University Seoul Korea Republic of Department of Applied Statistics Yonsei University Seoul Korea Republic of Department of Radiology College of Medicine Yonsei University Seoul Korea Republic of Department of Psychology Yonsei University Seoul Korea Republic of Cleveland Clinic United States

Following recent developments in quantum machine learning techniques, several algorithms have been developed for disease detection. This study explored the application of a hybrid quantum-classical algorithm for classifying region-of-interest time-series data obtained from resting-state functional magnetic resonance imaging (FMRI) in patients with early-stage cognitive impairment. Classical one-dimensional convolutional layers were used in conjunction with quantum convolutional neural networks in our hybrid algorithm. In a classical simulation, the proposed hybrid algorithms in our study exhibited higher balanced accuracies than classical convolutional neural networks under similar training conditions. In addition, in our study, among the 116 brain regions, two brain regions (the right hippocampus and left parahippocampus) were found to be relatively effective for classification based on the model performance. The associations of the two selected regions with cognitive decline, as found in previous studies, were validated using seed-based functional connectivity analysis. Thus, we confirmed both improvement in model performance with the quantum convolutional neural network and neuroscientific validity of brain regions from our hybrid quantum-classical model. © 2024, The Authors. All rights reserved.

关键词： Convolutional neural networks

Fdm: Effective and Efficient Incident Detection on Sparse Trajectory data

学校读者我要写书评

暂无评论

SSRN

SSRN 2023年

作者： Han, Xiaolin Grubenmann, Tobias Ma, Chenhao Li, Xiaodong Sun, Wenya Wong, Sze Chun Shang, Xuequn Cheng, Reynold The Northwestern Polytechnical University Xi’an China University of Southampton Southampton United Kingdom The Chinese University of Hong Kong Shenzhen China The University of Hong Kong Hong Kong HKU Musketeers Foundation Institute of Data Science Hong Kong Guangdong–Hong Kong-Macau Joint Laboratory Program 2020 China Department of Computer Science and Engineering Hong Kong School of Computer Science and Engineering China

Incident detection (ID), or the automatic discovery of anomalies from road traffic data (e.g., road sensor and GPS data), enables emergency actions (e.g., rescuing injured people) to be carried out in a timely fashion. Existing ID solutions based on data mining or machine learning often rely on dense traffic data;for instance, sensors installed in highways provide frequent updates of road information. In this paper, we ask the question: can ID be performed on sparse traffic data (e.g., location data obtained from GPS devices equipped on vehicles)? As these data may not be enough to describe the state of the roads involved, they can undermine the effectiveness of existing ID solutions. To tackle this challenge, we borrow an important insight from the transportation area, which uses trajectories (i.e., moving histories of vehicles) to derive incident patterns. We study how to obtain incident patterns from trajectories and devise a new solution (called Filter-Discovery-Match (FDM)) to detect anomalies in sparse traffic data. We have also developed a fast algorithm to support FDM. Experiments on a taxi dataset in Hong Kong and a simulated dataset show that FDM is more effective than state-of-the-art ID solutions on sparse traffic data, and is also efficient. © 2023, The Authors. All rights reserved.

关键词： Trajectories

DruSiLa: an integrated, in-silico disease similarity-based approach for drug repurposing

学校读者我要写书评

暂无评论

DruSiLa: an integrated, in-silico disease similarity-based a...

IEEE International Conference on Bioinformatics and Biomedicine (BIBM)

作者： Pratuat Amatya Paola Stolfi Flavio Lombardi Paolo Tieri Data Science Master Program Sapienza University of Rome Rome Italy IAC Institute for Applied Computing CNR National Research Council of Italy Rome Italy

ISBN: (数字)9781665468190

ISBN: (纸本)9781665468206

The importance of faster drug development has never been more evident than in present time when the whole world is struggling to cope up with the COVID-19 pandemic. At times when timely development of effective drugs and treatment plans could potentially save millions of lives, drug repurposing is one area of medicine that has garnered much of research interest. Apart from experimental drug repurposing studies that happen within wet labs, lot many new quantitative methods have been proposed in the literature. In this paper, one such quantitative methods for drug repurposing is implemented and evaluated. DruSiLa (DRUg in-SIlico LAboratory) is an in-silico drug repurposing method that leverages disease similarity measures to quantitatively rank existing drugs for their potential therapeutic efficacy against novel diseases. The proposed method makes use of available, manually curated, and open datasets on diseases, their genetic origins, and disease-related patho-phenotypes. DruSiLa evaluates pairwise disease similarity scores of any given target disease to each known disease in our dataset. Such similarity scores are then propagated through disease-drug associations, and aggregated at drug nodes to rank them for their predicted effectiveness against the target disease.

关键词： Drugs COVID-19 Pandemics Biomedical measurement Prediction algorithms Genetics Breast cancer