Search

CD4+T cells are a key mediator of various autoimmune diseases; however, how they contribute to disease development remains obscure primarily because of their cellular heterogeneity. Here, we evaluated CD4+T cell subpopulations by decomposition-based transcriptome characterization together with canonical clustering strategies. This approach identified 12 independent transcriptional gene programs governing whole CD4+T cell heterogeneity, which can explain the ambiguity of canonical clustering. In addition, we performed a meta-analysis using public single-cell data sets of over 1.8M peripheral CD4+T cells from 953 individuals by projecting cells onto the reference and cataloged cell frequency and qualitative alterations of the populations in 20 diseases. The analyses revealed that the 12 transcriptional programs were useful in characterizing each autoimmune disease and predicting its clinical status. Moreover, genetic variants associated with autoimmune diseases showed disease-specific enrichment within the 12 gene programs. The results collectively provide a landscape of single-cell transcriptomes of CD4+T cell subpopulations involved in autoimmune disease.

In order to incorporate a directive sound source into acoustic simulation using the finite-difference time-domain method (FDTD), this paper proposes an optimization-based method to estimate the initial value which approximates a desired directional pattern after propagation. The proposed method explicitly considers a discretized FDTD scheme and optimizes the initial value directly in the time domain so that every effect of the discretization error of FDTD, including numerical dispersion, is taken into account. It is also able to consider a frequency-wise directivity by integrating the Fourier transform into the optimization procedure, even though the estimated result is defined in the time domain. After the optimization, the obtained result can be utilized in any acoustic simulation based on the same FDTD scheme without modification because the result is represented as the initial value to be propagated and no additional procedure is required.

Inspired by the recent progress in self-supervised learning for computer vision that generates supervision using data augmentations, we explore a new general-purpose audio representation learning approach. We propose learning general-purpose audio representation from a single audio segment without expecting relationships between different time segments of audio samples. To implement this principle, we introduce Bootstrap Your Own Latent (BYOL) for Audio (BYOL-A, pronounced "viola"), an audio self-supervised learning method based on BYOL for learning general-purpose audio representation. Unlike most previous audio self-supervised learning methods that rely on agreement of vicinity audio segments or disagreement of remote ones, BYOL-A creates contrasts in an augmented audio segment pair derived from a single audio segment. With a combination of normalization and augmentation techniques, BYOL-A achieves state-of-the-art results in various downstream tasks. Extensive ablation studies also clarified the contribution of each component and their combinations., Comment: IJCNN 2021, 8 pages, 4 figures

The first total syntheses of JBIR-06 and two analogous depsipeptides, 12-membered antimycin-class antibiotics, have been accomplished via Shiina macrolactonization. Comparison of the spectroscopic data of the synthesized compounds with those reported for natural products verified that the absolute configutation of the natural products was (2 S,4 S,6 S,7 R,14 S).

We propose an end-to-end speech enhancement method with trainable time-frequency~(T-F) transform based on invertible deep neural network~(DNN). The resent development of speech enhancement is brought by using DNN. The ordinary DNN-based speech enhancement employs T-F transform, typically the short-time Fourier transform~(STFT), and estimates a T-F mask using DNN. On the other hand, some methods have considered end-to-end networks which directly estimate the enhanced signals without T-F transform. While end-to-end methods have shown promising results, they are black boxes and hard to understand. Therefore, some end-to-end methods used a DNN to learn the linear T-F transform which is much easier to understand. However, the learned transform may not have a property important for ordinary signal processing. In this paper, as the important property of the T-F transform, perfect reconstruction is considered. An invertible nonlinear T-F transform is constructed by DNNs and learned from data so that the obtained transform is perfectly reconstructing filterbank., To appear in Proc. IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP 2020)

Pakistan and northwestern India have frequently experienced severe heavy rainfall events during the boreal summer over the last 50 years including an event in late July and early August 2010 due to a sequence of monsoon surges. This study identified five dominant atmospheric patterns by applying principal component analysis and k-means clustering to a long-term sea level pressure dataset from 1979 to 2014. Two of these five dominant atmospheric patterns corresponded with a high frequency of the persistent atmospheric blocking index and positive sea level pressure over western Russia as well as an adjacent meridional trough ahead of northern Pakistan. In these two groups, a negative sea surface temperature anomaly was apparent over the equatorial mid- to eastern Pacific Ocean. The heavy precipitation periods with high persistent blocking frequency in western Russia as in the 2010 heat wave tended to have 1.2 times larger precipitation intensity compared to the whole of the heavy precipitation periods during the 36 years.

We propose a data-driven design method of perfect-reconstruction filterbank (PRFB) for sound-source enhancement (SSE) based on deep neural network (DNN). DNNs have been used to estimate a time-frequency (T-F) mask in the short-time Fourier transform (STFT) domain. Their training is more stable when a simple cost function as mean-squared error (MSE) is utilized comparing to some advanced cost such as objective sound quality assessments. However, such a simple cost function inherits strong assumptions on the statistics of the target and/or noise which is often not satisfied, and the mismatch of assumption results in degraded performance. In this paper, we propose to design the frequency scale of PRFB from training data so that the assumption on MSE is satisfied. For designing the frequency scale, the warped filterbank frame (WFBF) is considered as PRFB. The frequency characteristic of learned WFBF was in between STFT and the wavelet transform, and its effectiveness was confirmed by comparison with a standard STFT-based DNN whose input feature is compressed into the mel scale., Comment: 5 pages, to appear in IEEE ICASSP 2019 (Paper Code: AASP-P8.8, Session: Spatial Audio, Audio Enhancement and Bandwidth Extension)

Wave-based acoustic simulation methods are studied actively for predicting acoustical phenomena. Finite-difference time-domain (FDTD) method is one of the most popular methods owing to its straightforwardness of calculating an impulse response. In an FDTD simulation, an omnidirectional sound source is usually adopted, which is not realistic because the real sound sources often have specific directivities. However, there is very little research on imposing a directional sound source into FDTD methods. In this paper, a method of realizing a directional sound source in FDTD methods is proposed. It is formulated as an estimation problem of the initial value so that the estimated result corresponds to the desired directivity. The effectiveness of the proposed method is illustrated through some numerical experiments.