no code implementations • 26 Jan 2024 • Yeachan Park, Geonho Hwang, Wonyeol Lee, Sejun Park
In this work, we analyze the expressive power of neural networks under a more realistic setup: when we use floating-point numbers and operations.
no code implementations • 30 Aug 2023 • Geonho Hwang
By employing the aforementioned framework and the Whitney embedding theorem, we provide an upper bound for the minimum width, given by $\operatorname{max}(2d_x+1, d_y) + \alpha(\sigma)$, where $0 \leq \alpha(\sigma) \leq 2$ represents a constant depending on the activation function.
no code implementations • 25 Nov 2022 • Chang hoon Song, Geonho Hwang, Jun Ho Lee, Myungjoo Kang
In this study, we prove the universality of deep narrow RNNs and show that the upper bound of the minimum width for universality can be independent of the length of the data.
no code implementations • 18 Nov 2022 • Geonho Hwang, Myungjoo Kang
Despite its widespread adoption, our understanding of its universal approximation properties has been limited due to its intricate nature.
no code implementations • 11 Oct 2022 • Jaewoong Choi, Geonho Hwang, Hyunsoo Cho, Myungjoo Kang
This semantic basis represents sample-independent meaningful perturbations that change the same semantic attribute of an image on the entire latent space.
no code implementations • 26 May 2022 • Jaewoong Choi, Geonho Hwang, Hyunsoo Cho, Myungjoo Kang
In this paper, we approach this problem through a geometric analysis of latent spaces as a manifold.
2 code implementations • ICLR 2022 • Jaewoong Choi, Junho Lee, Changyeon Yoon, Jung Ho Park, Geonho Hwang, Myungjoo Kang
The global warpage implies that the latent space is not well-aligned globally and therefore the global traversal directions are bound to show limited success on it.
no code implementations • 17 Sep 2020 • Jaewoong Choi, Geonho Hwang, Myungjoo Kang
To represent these generative factors of data, we introduce two sets of continuous latent variables, private variable and public variable.