To enhance low-light images to normally-exposed ones is highly ill-posed, namely that the mapping relationship between them is one-to-many. Previous works based on the pixel-wise reconstruction losses and deterministic processes fail to capture the c
omplex conditional distribution of normally exposed images, which results in improper brightness, residual noise, and artifacts. In this paper, we investigate to model this one-to-many relationship via a proposed normalizing flow model. An invertible network that takes the low-light images/features as the condition and learns to map the distribution of normally exposed images into a Gaussian distribution. In this way, the conditional distribution of the normally exposed images can be well modeled, and the enhancement process, i.e., the other inference direction of the invertible network, is equivalent to being constrained by a loss function that better describes the manifold structure of natural images during the training. The experimental results on the existing benchmark datasets show our method achieves better quantitative and qualitative results, obtaining better-exposed illumination, less noise and artifact, and richer colors.
Cross-speaker style transfer (CSST) in text-to-speech (TTS) synthesis aims at transferring a speaking style to the synthesised speech in a target speakers voice. Most previous CSST approaches rely on expensive high-quality data carrying desired speak
ing style during training and require a reference utterance to obtain speaking style descriptors as conditioning on the generation of a new sentence. This work presents Referee, a robust reference-free CSST approach for expressive TTS, which fully leverages low-quality data to learn speaking styles from text. Referee is built by cascading a text-to-style (T2S) model with a style-to-wave (S2W) model. Phonetic PosteriorGram (PPG), phoneme-level pitch and energy contours are adopted as fine-grained speaking style descriptors, which are predicted from text using the T2S model. A novel pretrain-refinement method is adopted to learn a robust T2S model by only using readily accessible low-quality data. The S2W model is trained with high-quality target data, which is adopted to effectively aggregate style descriptors and generate high-fidelity speech in the target speakers voice. Experimental results are presented, showing that Referee outperforms a global-style-token (GST)-based baseline approach in CSST.
Denote by $tau$ k (n), $omega$(n) and $mu$ 2 (n) the number of representations of n as product of k natural numbers, the number of distinct prime factors of n and the characteristic function of the square-free integers, respectively. Let [t] be the i
ntegral part of real number t. For f = $omega$, 2 $omega$ , $mu$ 2 , $tau$ k , we prove that n x f x n = x d 1 f (d) d(d + 1) + O $epsilon$ (x $theta$ f +$epsilon$) for x $rightarrow$ $infty$, where $theta$ $omega$ = 53 110 , $theta$ 2 $omega$ = 9 19 , $theta$ $mu$2 = 2 5 , $theta$ $tau$ k = 5k--1 10k--1 and $epsilon$ > 0 is an arbitrarily small positive number. These improve the corresponding results of Bordell{`e}s.
Fascinating new phases of matter can emerge from strong electron interactions in solids. In recent years, a new exotic class of many-body phases, described by generalized electromagnetism of symmetric rank-2 electric and magnetic fields and immobile
charge excitations dubbed fractons, has attracted wide attention. Beside interesting properties in their own right, they are also closely related to gapped fracton quantum orders, new phases of dipole-coversing systems, quantum information, and quantum gravity. However, experimental realization of the rank-2 U(1) gauge theory is still absent, and even known practical experimental routes are scarce. In this work we propose a scheme of coupled optical phonons and nematics as well as several of its concrete experimental constructions. They can realize the electrostatics sector of the rank-2 U(1) gauge theory. A great advantage of our scheme is that it requires only basic ingredients of phonon and nematic physics, hence can be applied to a wide range of nematic matters from liquid crystals to electron orbitals. We expect this work will provide crucial guidance for the realization of rank-2 U(1) and fracton states of matter on a variety of platforms.
In this paper we propose a novel millimeter wave (mmW) multiple access method that exploits unique frequency dependent beamforming capabilities of True Time Delay (TTD) array architecture. The proposed protocol combines a contentionbased grant-free a
ccess and orthogonal frequency-division multiple access (OFDMA) scheme for uplink machine type communications. By exploiting abundant time-frequency resource blocks in mmW spectrum, we design a simple protocol that can achieve low collision rate and high network reliability for short packets and sporadic transmissions. We analyze the impact of various system parameters on system performance during synchronization and contention period. We exploit unique advantages of frequency dependent beamforming, referred as rainbow beam, to eliminate beam training overhead and analyze its impact on rates, latency, and coverage. The proposed system and protocol can flexibly accommodate different low latency applications with moderate rate requirements for a very large number of narrowband single antenna devices. By harnessing abundant resources in mmW spectrum and beamforming gain of TTD arrays rainbow link based system can simultaneously satisfy ultra-reliability and massive multiple access requirements.
There has been much interest in deploying deep learning algorithms on low-powered devices, including smartphones, drones, and medical sensors. However, full-scale deep neural networks are often too resource-intensive in terms of energy and storage. A
s a result, the bulk part of the machine learning operation is therefore often carried out on an edge server, where the data is compressed and transmitted. However, compressing data (such as images) leads to transmitting information irrelevant to the supervised task. Another popular approach is to split the deep network between the device and the server while compressing intermediate features. To date, however, such split computing strategies have barely outperformed the aforementioned naive data compression baselines due to their inefficient approaches to feature compression. This paper adopts ideas from knowledge distillation and neural image compression to compress intermediate feature representations more efficiently. Our supervised compression approach uses a teacher model and a student model with a stochastic bottleneck and learnable prior for entropy coding. We compare our approach to various neural image and feature compression baselines in three vision tasks and found that it achieves better supervised rate-distortion performance while also maintaining smaller end-to-end latency. We furthermore show that the learned feature representations can be tuned to serve multiple downstream tasks.
In this paper, we present a self-training method, named ST3D++, with a holistic pseudo label denoising pipeline for unsupervised domain adaptation on 3D object detection. ST3D++ aims at reducing noise in pseudo label generation as well as alleviating
the negative impacts of noisy pseudo labels on model training. First, ST3D++ pre-trains the 3D object detector on the labeled source domain with random object scaling (ROS) which is designed to reduce target domain pseudo label noise arising from object scale bias of the source domain. Then, the detector is progressively improved through alternating between generating pseudo labels and training the object detector with pseudo-labeled target domain data. Here, we equip the pseudo label generation process with a hybrid quality-aware triplet memory to improve the quality and stability of generated pseudo labels. Meanwhile, in the model training stage, we propose a source data assisted training strategy and a curriculum data augmentation policy to effectively rectify noisy gradient directions and avoid model over-fitting to noisy pseudo labeled data. These specific designs enable the detector to be trained on meticulously refined pseudo labeled target data with denoised training signals, and thus effectively facilitate adapting an object detector to a target domain without requiring annotations. Finally, our method is assessed on four 3D benchmark datasets (i.e., Waymo, KITTI, Lyft, and nuScenes) for three common categories (i.e., car, pedestrian and bicycle). ST3D++ achieves state-of-the-art performance on all evaluated settings, outperforming the corresponding baseline by a large margin (e.g., 9.6% $sim$ 38.16% on Waymo $rightarrow$ KITTI in terms of AP$_{text{3D}}$), and even surpasses the fully supervised oracle results on the KITTI 3D object detection benchmark with target prior. Code will be available.
In recent years, the field of antiferromagnetic spintronics has been substantially advanced. Electric-field control is a promising approach to achieving ultra-low power spintronic devices via suppressing Joule heating. In this article, cutting-edge r
esearch, including electric-field modulation of antiferromagnetic spintronic devices using strain, ionic liquids, dielectric materials, and electrochemical ionic migration, are comprehensively reviewed. Various emergent topics such as the Neel spin-orbit torque, chiral spintronics, topological antiferromagnetic spintronics, anisotropic magnetoresistance, memory devices, two-dimensional magnetism, and magneto-ionic modulation with respect to antiferromagnets are examined. In conclusion, we envision the possibility of realizing high-quality room-temperature antiferromagnetic tunnel junctions, antiferromagnetic spin logic devices, and artificial antiferromagnetic neurons. It is expected that this work provides an appropriate and forward-looking perspective that will promote the rapid development of this field.
One of the main bottleneck issues for room-temperature antiferromagnetic spintronic devices is the small signal read-out owing to the limited anisotropic magnetoresistance in antiferromagnets. However, this could be overcome by either utilizing the B
erry-curvature-induced anomalous Hall resistance in noncollinear antiferromagnets or establishing tunnel junction devices based on effective manipulation of antiferromagnetic spins. In this work, we demonstrate the giant piezoelectric strain control of the spin structure and the anomalous Hall resistance in a noncollinear antiferromagnetic metal - D019 hexagonal Mn3Ga. Furthermore, we built tunnel junction devices with a diameter of 200 nm to amplify the maximum tunneling resistance ratio to more than 10% at room-temperature, which thus implies significant potential of noncollinear antiferromagnets for large signal-output and high-density antiferromagnetic spintronic device applications.
Online sexism has become an increasing concern in social media platforms as it has affected the healthy development of the Internet and can have negative effects in society. While research in the sexism detection domain is growing, most of this resea
rch focuses on English as the language and on Twitter as the platform. Our objective here is to broaden the scope of this research by considering the Chinese language on Sina Weibo. We propose the first Chinese sexism dataset -- Sina Weibo Sexism Review (SWSR) dataset --, as well as a large Chinese lexicon SexHateLex made of abusive and gender-related terms. We introduce our data collection and annotation process, and provide an exploratory analysis of the dataset characteristics to validate its quality and to show how sexism is manifested in Chinese. The SWSR dataset provides labels at different levels of granularity including (i) sexism or non-sexism, (ii) sexism category and (iii) target type, which can be exploited, among others, for building computational methods to identify and investigate finer-grained gender-related abusive language. We conduct experiments for the three sexism classification tasks making use of state-of-the-art machine learning models. Our results show competitive performance, providing a benchmark for sexism detection in the Chinese language, as well as an error analysis highlighting open challenges needing more research in Chinese NLP. The SWSR dataset and SexHateLex lexicon are publicly available.
