arXiv subjects

Stefan Bamberger

Publications and source records attributed to Stefan Bamberger.

4 recordsLinked to original sources

arXiv2023

Approximating Positive Homogeneous Functions with Scale Invariant Neural Networks

We investigate to what extent it is possible to solve linear inverse problems with $ReLu$ networks. Due to the scaling invariance arising from the linearity, an optimal reconstruction function $f$ for such a problem is positive homogeneous, i.e., satisfies $f(λx) = λf(x)$ for all non-negative $λ$. In a $ReLu$ network, this condition translates to considering networks without bias terms. We first consider recovery of sparse vectors from few linear measurements. We prove that $ReLu$- networks with only one hidden layer cannot even recover $1$-sparse vectors, not even approximately, and regardless of the width of the network. However, with two hidden layers, approximate recovery with arbitrary precision and arbitrary sparsity level $s$ is possible in a stable way. We then extend our results to a wider class of recovery problems including low-rank matrix recovery and phase retrieval. Furthermore, we also consider the approximation of general positive homogeneous functions with neural networks. Extending previous work, we establish new results explaining under which conditions such functions can be approximated with neural networks. Our results also shed some light on the seeming contradiction between previous works showing that neural networks for inverse problems typically have very large Lipschitz constants, but still perform very well also for adversarial noise. Namely, the error bounds in our expressivity results include a combination of a small constant term and a term that is linear in the noise level, indicating that robustness issues may occur only for very small noise levels.

cs.LG↗

arXiv2021

The Hanson-Wright Inequality for Random Tensors

We provide moment bounds for expressions of the type $(X^{(1)} \otimes \dots \otimes X^{(d)})^T A (X^{(1)} \otimes \dots \otimes X^{(d)})$ where $\otimes$ denotes the Kronecker product and $X^{(1)}, \dots, X^{(d)}$ are random vectors with independent, mean 0, variance 1, subgaussian entries. The bounds are tight up to constants depending on $d$ for the case of Gaussian random vectors. Our proof also provides a decoupling inequality for expressions of this type. Using these bounds, we obtain new, improved concentration inequalities for expressions of the form $\|B (X^{(1)} \otimes \dots \otimes X^{(d)})\|_2$.

math.PR↗

arXiv2021

Johnson-Lindenstrauss Embeddings with Kronecker Structure

We prove the Johnson-Lindenstrauss property for matrices $ΦD_ξ$ where $Φ$ has the restricted isometry property and $D_ξ$ is a diagonal matrix containing the entries of a Kronecker product $ξ= ξ^{(1)} \otimes \dots \otimes ξ^{(d)}$ of $d$ independent Rademacher vectors. Such embeddings have been proposed in recent works for a number of applications concerning compression of tensor structured data, including the oblivious sketching procedure by Ahle et al. for approximate tensor computations. For preserving the norms of $p$ points simultaneously, our result requires $Φ$ to have the restricted isometry property for sparsity $C(d) (\log p)^d$. In the case of subsampled Hadamard matrices, this can improve the dependence of the embedding dimension on $p$ to $(\log p)^d$ while the best previously known result required $(\log p)^{d + 1}$. That is, for the case of $d=2$ at the core of the oblivious sketching procedure by Ahle et al., the scaling improves from cubic to quadratic. We provide a counterexample to prove that the scaling established in our result is optimal under mild assumptions.

cs.DS↗

arXiv2020

Optimal Fast Johnson-Lindenstrauss Embeddings for Large Data Sets

Johnson-Lindenstrauss embeddings are widely used to reduce the dimension and thus the processing time of data. To reduce the total complexity, also fast algorithms for applying these embeddings are necessary. To date, such fast algorithms are only available either for a non-optimal embedding dimension or up to a certain threshold on the number of data points. We address a variant of this problem where one aims to simultaneously embed larger subsets of the data set. Our method follows an approach by Nelson: A subsampled Hadamard transform maps points into a space of lower, but not optimal dimension. Subsequently, a random matrix with independent entries projects to an optimal embedding dimension. For subsets whose size scales at least polynomially in the ambient dimension, the complexity of this method comes close to the number of operations just to read the data under mild assumptions on the size of the data set that are considerably less restrictive than in previous works. We also prove a lower bound showing that subsampled Hadamard matrices alone cannot reach an optimal embedding dimension. Hence, the second embedding cannot be omitted.

cs.DS↗