Search results
25 de ene. de 2024 · A survey of deep learning and foundation models for time series forecasting, by John A. Miller and others. The paper reviews state-of-the-art techniques and suggests future work, with a DOI and submission history.
- [1805.10369] Stable Recurrent Models
John Miller, Moritz Hardt. Stability is a fundamental...
- [1805.10369] Stable Recurrent Models
25 de may. de 2018 · John Miller and Moritz Hardt investigate the stability of recurrent neural networks and their approximation by feed-forward networks. The paper, to appear in ICLR 2019, shows that stable recurrent models can perform well on sequence tasks.
My primary research interests are in Bayesian optimization, transfer learning, Bayesian non-parametrics, and statistical and machine learning tools for physics/engineering and epidemiology. I am an Alfred P. Sloan 2021-22 Minority Scholar at Duke.
Statistical Methods for Physics and Engineering, Transfer Learning, Gaussian Processes, and Bayesian Inference. John Miller | Scholars@Duke profile: Research.
A paper on arXiv that investigates stability and approximation of recurrent neural networks. The authors prove stable recurrent models can be approximated by feed-forward networks and show stable models often perform well on sequence tasks.
20 de jun. de 2019 · An explicit combinatorial minimal free resolution of an arbitrary monomial ideal I in a polynomial ring in n variables over a field of characteristic 0 is defined canonically, without any choices, using higher-dimensional generalizations of combined spanning trees for cycles and cocycles ("hedges") in the upper Koszul simplicial complexes of I a...
redict facts happening in the future. This paper proposes Recurrent Event Network (RE-NET), a novel autoregressive architectu. e for predicting future interactions. The occur-rence of a fact (event) is modeled as a proba-bility distribution conditioned on tempora.
