Search

In this paper, we investigate solving the elliptical multiple eigenvalue (EME) problems using a Feedforward Neural Network. Firstly, we propose a general formulation for computing EME based on penalized variational forms of elliptical eigenvalue problems. Next, we solve the penalized variational form using the Deep Ritz Method. We establish an upper bound on the error between the estimated eigenvalues and true ones in terms of the depth D , width W of the neural network, and training sample size n. By exploring the regularity of the EME and selecting an appropriate depth D and width W , we demonstrate that the desired bound enjoys a convergence rate of O (1 / n 16) , which circumvents the curse of dimensionality. We also present several high-dimensional simulation results to illustrate the effectiveness of our proposed method and support our theoretical findings. [ABSTRACT FROM AUTHOR]

In this paper, we consider recovering n - dimensional signals from m binary measurements corrupted by noises and sign flips under the assumption that the target signals have low generative intrinsic dimension, i.e., the target signals can be approximately generated via an L-Lipschitz generator G : R k → R n , k ≪ n . Although the binary measurements model is highly nonlinear, we propose a least square decoder and prove that, up to a constant c, with high probability, the least square decoder achieves a sharp estimation error C k log (L n) m as long as m ≥ C (k log (L n)) . Extensive numerical simulations and comparisons with state-of-the-art methods demonstrated the least square decoder is robust to noise and sign flips, as indicated by our theory. By constructing a ReLU network with properly chosen depth and width, we verify the (approximately) deep generative prior, which is of independent interest. [ABSTRACT FROM AUTHOR]

We propose a fast Newton algorithm for ℓ 0 regularized high-dimensional generalized linear models based on support detection and root finding. We refer to the proposed method as GSDAR. GSDAR is developed based on the KKT conditions for ℓ 0 -penalized maximum likelihood estimators and generates a sequence of solutions of the KKT system iteratively. We show that GSDAR can be equivalently formulated as a generalized Newton algorithm. Under a restricted invertibility condition on the likelihood function and a sparsity condition on the regression coefficient, we establish an explicit upper bound on the estimation errors of the solution sequence generated by GSDAR in supremum norm and show that it achieves the optimal order in finite iterations with high probability. Moreover, we show that the oracle estimator can be recovered with high probability if the target signal is above the detectable level. These results directly concern the solution sequence generated from the GSDAR algorithm, instead of a theoretically defined global solution. We conduct simulations and real data analysis to illustrate the effectiveness of the proposed method. [ABSTRACT FROM AUTHOR]