Search

Deep neural networks have been applied to improve the image quality of fluorescence microscopy imaging. Previous methods are based on convolutional neural networks (CNNs) which generally require more time-consuming training of separate models for each new imaging experiment, impairing the applicability and generalization. Once the model is trained (typically with tens to hundreds of image pairs) it can then be used to enhance new images that are like the training data. In this study, we proposed a novel imaging-transformer based model, Convolutional Neural Network Transformer (CNNT), to outperform the CNN networks for image denoising. In our scheme we have trained a single CNNT based backbone model from pairwise high-low SNR images for one type of fluorescence microscope (instance structured illumination, iSim). Fast adaption to new applications was achieved by fine-tuning the backbone on only 5-10 sample pairs per new experiment. Results show the CNNT backbone and fine-tuning scheme significantly reduces the training time and improves the image quality, outperformed training separate models using CNN approaches such as - RCAN and Noise2Fast. Here we show three examples of the efficacy of this approach on denoising wide-field, two-photon and confocal fluorescence data. In the confocal experiment, which is a 5 by 5 tiled acquisition, the fine-tuned CNNT model reduces the scan time form one hour to eight minutes, with improved quality.

As artificial intelligence (AI) is integrated into various services and systems in society, many companies and organizations have proposed AI principles, policies, and made the related commitments. Conversely, some have proposed the need for independent audits, arguing that the voluntary principles adopted by the developers and providers of AI services and systems insufficiently address risk. This policy recommendation summarizes the issues related to the auditing of AI services and systems and presents three recommendations for promoting AI auditing that contribute to sound AI governance. Recommendation1.Development of institutional design for AI audits. Recommendation2.Training human resources for AI audits. Recommendation3. Updating AI audits in accordance with technological progress. In this policy recommendation, AI is assumed to be that which recognizes and predicts data with the last chapter outlining how generative AI should be audited.

Feigin-Frenkel duality is the isomorphism between the principal $\mathcal{W}$-algebras of a simple Lie algebra $\mathfrak{g}$ and its Langlands dual Lie algebra ${}^L\mathfrak{g}$. A generalization of this duality to a larger family of $\mathcal{W}$-algebras called hook-type was recently conjectured by Gaiotto and Rap\v{c}\'ak and proved by the first two authors. It says that the affine cosets of two different hook-type $\mathcal{W}$-algebras are isomorphic. A natural question is whether the duality between affine cosets can be enhanced to a duality between the full $\mathcal{W}$-algebras. There is a convolution operation that maps a hook-type $\mathcal{W}$-algebra $\mathcal{W}$ to a certain relative semi-infinite cohomology of $\mathcal{W}$ tensored with a suitable kernel VOA. The first two authors conjectured previously that this cohomology is isomorphic to the Feigin-Frenkel dual hook-type $\mathcal{W}$-algebra. Our main result is a proof of this conjecture., Comment: 19 pages

Although application examples of multilevel optimization have already been discussed since the 1990s, the development of solution methods was almost limited to bilevel cases due to the difficulty of the problem. In recent years, in machine learning, Franceschi et al. have proposed a method for solving bilevel optimization problems by replacing their lower-level problems with the $T$ steepest descent update equations with some prechosen iteration number $T$. In this paper, we have developed a gradient-based algorithm for multilevel optimization with $n$ levels based on their idea and proved that our reformulation asymptotically converges to the original multilevel problem. As far as we know, this is one of the first algorithms with some theoretical guarantee for multilevel optimization. Numerical experiments show that a trilevel hyperparameter learning model considering data poisoning produces more stable prediction results than an existing bilevel hyperparameter learning model in noisy data settings., Comment: NeurIPS 2021 camera-ready, 27 pages

Based on the Kazama-Suzuki type coset construction and its inverse coset between the subregular $\mathcal{W}$-algebras for $\mathfrak{sl}_n$ and the principal $\mathcal{W}$-superalgebras for $\mathfrak{sl}_{1|n}$, we prove weight-wise linear equivalences of their representation categories. Our main results are then improvements of these correspondences incorporating the monoidal structures. Firstly, in the rational case, we obtain the classification of simple modules and their fusion rules via simple current extensions from their Heisenberg cosets. Secondly, beyond the rational case, we use certain kernel VOAs together with relative semi-infinite cohomology functors to get functors from categories of modules for the subregular $\mathcal{W}$-algebras for $\mathfrak{sl}_n$ to categories of modules for the principal $\mathcal{W}$-superalgebras for $\mathfrak{sl}_{1|n}$ and vice versa. We study these functors and in particular prove isomorphisms between the superspaces of logarithmic intertwining operators. As a corollary, we obtain correspondences of representation categories in the monoidal sense beyond the rational case as well., Comment: 51 pages

We study the representation theory of the Kazama-Suzuki coset vertex operator superalgebra associated with the pair of a complex simple Lie algebra and its Cartan subalgebra. In the case of type $A_{1}$, B.L. Feigin, A.M. Semikhatov, and I.Yu. Tipunin introduced another coset construction, which is "inverse" of the Kazama-Suzuki coset construction. In this paper we generalize the latter coset construction to arbitrary type and establish a categorical equivalence between the categories of certain modules over an affine vertex operator algebra and the corresponding Kazama-Suzuki coset vertex operator superalgebra. Moreover, when the affine vertex operator algebra is regular, we prove that the corresponding Kazama-Suzuki coset vertex operator superalgebra is also regular and the category of its ordinary modules carries a braided monoidal category structure by the theory of vertex tensor categories., Comment: 29 pages, comments are welcome

In this paper we construct Bernstein--Gelfand--Gelfand type resolution of simple highest weight modules over the simple $\mathcal{N}=2$ vertex operator superalgebra of central charge $c_{p,p'}=3\left(1-\frac{2p'}{p}\right)$ by means of the Kazama--Suzuki coset construction. As an application, we compute the twisted Zhu algebras of the simple $\mathcal{N}=2$ vertex operator superalgebra. We also compute the Frenkel--Zhu bimodule structure associated with a certain simple highest weight module of central charge $c_{3,2}=-1$., Comment: 18 pages

We compute the modular transformation formula of the characters for a certain family of (finitely or uncountably many) simple modules over the simple $\mathcal{N}=2$ vertex operator superalgebra of central charge $c_{p,p'}=3\left(1-\frac{2p'}{p}\right),$ where $(p,p')$ is a pair of coprime positive integers such that $p\geq2$. When $p'=1$, the formula coincides with that of the $\mathcal{N}=2$ unitary minimal series found by F. Ravanini and S.-K. Yang. In addition, we study the properties of the corresponding "modular $S$-matrix", which is no longer a matrix if $p'\geq2$., Comment: 24 pages, definition of Verlinde coefficient in Section 5 is corrected, minor change in Proposition 3.8, references added