Search

Recent advances in generative models trained on large-scale datasets have made it possible to synthesize high-quality samples across various domains. Moreover, the emergence of strong inversion networks enables not only a reconstruction of real-world images but also the modification of attributes through various editing methods. However, in certain domains related to privacy issues, e.g., human faces, advanced generative models along with strong inversion methods can lead to potential misuses. In this paper, we propose an essential yet under-explored task called generative identity unlearning, which steers the model not to generate an image of a specific identity. In the generative identity unlearning, we target the following objectives: (i) preventing the generation of images with a certain identity, and (ii) preserving the overall quality of the generative model. To satisfy these goals, we propose a novel framework, Generative Unlearning for Any Identity (GUIDE), which prevents the reconstruction of a specific identity by unlearning the generator with only a single image. GUIDE consists of two parts: (i) finding a target point for optimization that un-identifies the source latent code and (ii) novel loss functions that facilitate the unlearning procedure while less affecting the learned distribution. Our extensive experiments demonstrate that our proposed method achieves state-of-the-art performance in the generative machine unlearning task. The code is available at https://github.com/KHU-AGI/GUIDE., Comment: 15 pages, 17 figures, 10 tables, CVPR 2024 Poster

We address a challenging lifelong few-shot image generation task for the first time. In this situation, a generative model learns a sequence of tasks using only a few samples per task. Consequently, the learned model encounters both catastrophic forgetting and overfitting problems at a time. Existing studies on lifelong GANs have proposed modulation-based methods to prevent catastrophic forgetting. However, they require considerable additional parameters and cannot generate high-fidelity and diverse images from limited data. On the other hand, the existing few-shot GANs suffer from severe catastrophic forgetting when learning multiple tasks. To alleviate these issues, we propose a framework called Lifelong Few-Shot GAN (LFS-GAN) that can generate high-quality and diverse images in lifelong few-shot image generation task. Our proposed framework learns each task using an efficient task-specific modulator - Learnable Factorized Tensor (LeFT). LeFT is rank-constrained and has a rich representation ability due to its unique reconstruction technique. Furthermore, we propose a novel mode seeking loss to improve the diversity of our model in low-data circumstances. Extensive experiments demonstrate that the proposed LFS-GAN can generate high-fidelity and diverse images without any forgetting and mode collapse in various domains, achieving state-of-the-art in lifelong few-shot image generation task. Surprisingly, we find that our LFS-GAN even outperforms the existing few-shot GANs in the few-shot image generation task. The code is available at Github., Comment: 20 pages, 19 figures, 14 tables, ICCV 2023 Poster

Opportunities for stochastic arbitrage in an options market arise when it is possible to construct a portfolio of options which provides a positive option premium and which, when combined with a direct investment in the underlying asset, generates a payoff which stochastically dominates the payoff from the direct investment in the underlying asset. We provide linear and mixed-integer linear programs for computing the stochastic arbitrage opportunity providing the maximum option premium to an investor. We apply our programs to 18 years of data on monthly put and call options on the Standard & Poors 500 index, finding no evidence that stochastic arbitrage opportunities are systematically present. A skewed specification of the underlying market return distribution with a constant market risk premium and constant multiplicative variance risk premium is broadly consistent with the pricing of market index options at moderate strikes., Comment: 22 pages, 7 figures, 7 tables

New nonparametric tests of copula exchangeability and radial symmetry are proposed. The novel aspect of the tests is a resampling procedure that exploits group invariance conditions associated with the relevant symmetry hypothesis. They may be viewed as feasible versions of randomization tests of symmetry, the latter being inapplicable due to the unobservability of margins. Our tests are simple to compute, control size asymptotically, consistently detect arbitrary forms of asymmetry, and do not require the specification of a tuning parameter. Simulations indicate excellent small sample properties compared to existing procedures involving the multiplier bootstrap.

This paper develops a uniformly valid and asymptotically nonconservative test based on projection for a class of shape restrictions. The key insight we exploit is that these restrictions form convex cones, a simple and yet elegant structure that has been barely harnessed in the literature. Based on a monotonicity property afforded by such a geometric structure, we construct a bootstrap procedure that, unlike many studies in nonstandard settings, dispenses with estimation of local parameter spaces, and the critical values are obtained in a way as simple as computing the test statistic. Moreover, by appealing to strong approximations, our framework accommodates nonparametric regression models as well as distributional/density-related and structural settings. Since the test entails a tuning parameter (due to the nonstandard nature of the problem), we propose a data-driven choice and prove its validity. Monte Carlo simulations confirm that our test works well., Comment: This version contains the following sections omitted from the published version: i) discussions of the examples in the main text, ii) proofs for Appendix C (in the online appendix), and iii) the complete set of simulation results. A previous version of this paper was circulated under the title "A General Framework for Inference on Shape Restrictions."

We develop a new statistical procedure to test whether the dependence structure is identical between two groups. Rather than relying on a single index such as Pearson's correlation coefficient or Kendall's Tau, we consider the entire dependence structure by investigating the dependence functions (copulas). The critical values are obtained by a modified randomization procedure designed to exploit asymptotic group invariance conditions. Implementation of the test is intuitive and simple, and does not require any specification of a tuning parameter or weight function. At the same time, the test exhibits excellent finite sample performance, with the null rejection rates almost equal to the nominal level even when the sample size is extremely small. Two empirical applications concerning the dependence between income and consumption, and the Brexit effect on European financial market integration are provided.