Search

To directly detect exoplanets and protoplanetary disks, the development of high accuracy wavefront sensing and control (WFS&C) technologies is essential, especially for ground-based Extremely Large Telescopes (ELTs). The Subaru Coronagraphic Extreme Adaptive Optics (SCExAO) instrument is a high-contrast imaging platform to discover and characterize exoplanets and protoplanetary disks. It also serves as a testbed to validate and deploy new concepts or algorithms for high-contrast imaging approaches for ELTs, using the latest hardware and software technologies on an 8-meter class telescope. SCExAO is a multi-band instrument, using light from 600 to 2500 nm, and delivering a high Strehl ratio (>80% in median seeing in H-band) downstream of a low-order correction provided by the facility AO188. Science observations are performed with coronagraphs, an integral field spectrograph, or single aperture interferometers. The SCExAO project continuously reaches out to the community for development and upgrades. Existing operating testbeds such as the SCExAO are also unique opportunities to test and deploy the new technologies for future ELTs. We present and show a live demonstration of the SCExAO capabilities (Real-time predictive AO control, Focal plane WFS&C, etc) as a host testbed for the remote collaborators to test and deploy the new WFS&C concepts or algorithms. We also present several high-contrast imaging technologies that are under development or that have already been demonstrated on-sky., Comment: 13 pages, 8 figures, 2021 SPIE Optics+Photonics

We introduced a measurement procedure for the involuntary response of voice fundamental-frequency to frequency modulated auditory stimulation. This involuntary response plays an essential role in voice fundamental frequency control while less investigated due to technical difficulties. This article introduces an interactive and real-time tool for investigating this response and supporting tools adopting our new measurement method. The method enables simultaneous measurement of multiple system properties based on a novel set of extended time-stretched pulses combined with orthogonalization. We made MATLAB implementation of these tools available as an open-source repository. This article also provides the detailed measurement procedure using the interactive tool followed by offline measurement tools for conducting subjective experiments and statistical analyses. It also provides technical descriptions of constituent signal processing subsystems as appendices. This application serves as an example for adopting our method to biological system analysis., Comment: Accepted for APSIPA ASC 2021

This article describes an efficient end-to-end speech translation (E2E-ST) framework based on non-autoregressive (NAR) models. End-to-end speech translation models have several advantages over traditional cascade systems such as inference latency reduction. However, conventional AR decoding methods are not fast enough because each token is generated incrementally. NAR models, however, can accelerate the decoding speed by generating multiple tokens in parallel on the basis of the token-wise conditional independence assumption. We propose a unified NAR E2E-ST framework called Orthros, which has an NAR decoder and an auxiliary shallow AR decoder on top of the shared encoder. The auxiliary shallow AR decoder selects the best hypothesis by rescoring multiple candidates generated from the NAR decoder in parallel (parallel AR rescoring). We adopt conditional masked language model (CMLM) and a connectionist temporal classification (CTC)-based model as NAR decoders for Orthros, referred to as Orthros-CMLM and Orthros-CTC, respectively. We also propose two training methods to enhance the CMLM decoder. Experimental evaluations on three benchmark datasets with six language directions demonstrated that Orthros achieved large improvements in translation quality with a very small overhead compared with the baseline NAR model. Moreover, the Conformer encoder architecture enabled large quality improvements, especially for CTC-based models. Orthros-CTC with the Conformer encoder increased decoding speed by 3.63x on CPU with translation quality comparable to that of an AR model.

In order to obtain insights into dynamics of developed plane Couette turbulence, this paper considers bifurcation structure of unstable periodic orbits (UPOs) in the large-eddy-simulation (LES) system with the Smagorinsky-type eddy viscosity model. Treating the Smagorinsky constant as a bifurcation parameter, we detect the bifurcation points connecting two known UPOs which were separately discovered in the Navier--Stokes system [G. Kawahara and S. Kida, J. Fluid Mech., 449, 291--300 (2001)]. At the moderately high Reynolds number, the LES UPO of the present study possesses the spanwise vortices which seem to be caused by the streak instability and appear in the central region of the channel. We note that to our knowledge a spanwise vortex has not been reported through the UPO analysis of wall flows. The stretched spanwise vortices of the LES UPO enhance transfer of the streamwise turbulent momentum as in developed near-wall turbulence.

Twisted bilayer graphene (TBLG) has emerged as an exciting new material with tunable electronic properties ranging from superconductivity to correlated insulating phases. But current methods of fabrication and identification of TBLG are painstaking and laborious. In this work, we combine Raman spectroscopy with the Gaussian mixture model (GMM) data clustering algorithm to identify areas with particular twist angles, from a TBLG sample with a mixture of orientations. We present two approaches to this cluster analysis: training the GMM on Raman parameters returned by peak fits, and on full Raman spectra with dimensionality reduced by principal component analysis. In both cases we demonstrate that GMM can identify regions of distinct twist angle from within Raman datacubes. We also show that once a model has been trained, and the identified clusters labelled, the model can be reapplied to new Raman scans to assess the similarity between the materials in the new region and the testing region. This could enable high-throughput fabrication of TBLG, by allowing computerised detection of particular twist angles from automated large-area scans.

In this work, we propose novel decoding algorithms to enable streaming automatic speech recognition (ASR) on unsegmented long-form recordings without voice activity detection (VAD), based on monotonic chunkwise attention (MoChA) with an auxiliary connectionist temporal classification (CTC) objective. We propose a block-synchronous beam search decoding to take advantage of efficient batched output-synchronous and low-latency input-synchronous searches. We also propose a VAD-free inference algorithm that leverages CTC probabilities to determine a suitable timing to reset the model states to tackle the vulnerability to long-form data. Experimental evaluations demonstrate that the block-synchronous decoding achieves comparable accuracy to the label-synchronous one. Moreover, the VAD-free inference can recognize long-form speech robustly for up to a few hours., Comment: Accepted at Interspeech 2021

Extracting the interaction rules of biological agents from movement sequences pose challenges in various domains. Granger causality is a practical framework for analyzing the interactions from observed time-series data; however, this framework ignores the structures and assumptions of the generative process in animal behaviors, which may lead to interpretational problems and sometimes erroneous assessments of causality. In this paper, we propose a new framework for learning Granger causality from multi-animal trajectories via augmented theory-based behavioral models with interpretable data-driven models. We adopt an approach for augmenting incomplete multi-agent behavioral models described by time-varying dynamical systems with neural networks. For efficient and interpretable learning, our model leverages theory-based architectures separating navigation and motion processes, and the theory-guided regularization for reliable behavioral modeling. This can provide interpretable signs of Granger-causal effects over time, i.e., when specific others cause the approach or separation. In experiments using synthetic datasets, our method achieved better performance than various baselines. We then analyzed multi-animal datasets of mice, flies, birds, and bats, which verified our method and obtained novel biological insights., Comment: 24 pages, 5 figures, to appear in NeurIPS 2021

While attention-based encoder-decoder (AED) models have been successfully extended to the online variants for streaming automatic speech recognition (ASR), such as monotonic chunkwise attention (MoChA), the models still have a large label emission latency because of the unconstrained end-to-end training objective. Previous works tackled this problem by leveraging alignment information to control the timing to emit tokens during training. In this work, we propose a simple alignment-free regularization method, StableEmit, to encourage MoChA to emit tokens earlier. StableEmit discounts the selection probabilities in hard monotonic attention for token boundary detection by a constant factor and regularizes them to recover the total attention mass during training. As a result, the scale of the selection probabilities is increased, and the values can reach a threshold for token emission earlier, leading to a reduction of emission latency and deletion errors. Moreover, StableEmit can be combined with methods that constraint alignments to further improve the accuracy and latency. Experimental evaluations with LSTM and Conformer encoders demonstrate that StableEmit significantly reduces the recognition errors and the emission latency simultaneously. We also show that the use of alignment information is complementary in both metrics., Comment: Accepted at Interspeech 2021