arXiv subjects

Qi Song

Publications and source records attributed to Qi Song.

At least 73 records · Page 4Linked to original sources

arXiv2024

BFRFormer: Transformer-based generator for Real-World Blind Face Restoration

Blind face restoration is a challenging task due to the unknown and complex degradation. Although face prior-based methods and reference-based methods have recently demonstrated high-quality results, the restored images tend to contain over-smoothed results and lose identity-preserved details when the degradation is severe. It is observed that this is attributed to short-range dependencies, the intrinsic limitation of convolutional neural networks. To model long-range dependencies, we propose a Transformer-based blind face restoration method, named BFRFormer, to reconstruct images with more identity-preserved details in an end-to-end manner. In BFRFormer, to remove blocking artifacts, the wavelet discriminator and aggregated attention module are developed, and spectral normalization and balanced consistency regulation are adaptively applied to address the training instability and over-fitting problem, respectively. Extensive experiments show that our method outperforms state-of-the-art methods on a synthetic dataset and four real-world datasets. The source code, Casia-Test dataset, and pre-trained models are released at https://github.com/s8Znk/BFRFormer.

cs.CV↗

arXiv2024

An Empirical Study on the Impact of Positional Encoding in Transformer-based Monaural Speech Enhancement

Transformer architecture has enabled recent progress in speech enhancement. Since Transformers are position-agostic, positional encoding is the de facto standard component used to enable Transformers to distinguish the order of elements in a sequence. However, it remains unclear how positional encoding exactly impacts speech enhancement based on Transformer architectures. In this paper, we perform a comprehensive empirical study evaluating five positional encoding methods, i.e., Sinusoidal and learned absolute position embedding (APE), T5-RPE, KERPLE, as well as the Transformer without positional encoding (No-Pos), across both causal and noncausal configurations. We conduct extensive speech enhancement experiments, involving spectral mapping and masking methods. Our findings establish that positional encoding is not quite helpful for the models in a causal configuration, which indicates that causal attention may implicitly incorporate position information. In a noncausal configuration, the models significantly benefit from the use of positional encoding. In addition, we find that among the four position embeddings, relative position embeddings outperform APEs.

eess.AS↗

arXiv2023

Enhancing carrier transfer properties of Na-rich anti-perovskites, Na4OM2 with tetrahedral anion groups: an evaluation through first-principles computational analysis

The practical application of Na-based solid-state electrolytes (SSEs) is limited by their low level of conduction. To evaluate the impact of tetrahedral anion groups on carrier migration, we designed a set of anti-perovskite SSEs theoretically based on the previously reported Na4OBr2, including Na4O(BH4)2, Na4O(BF4)2, and Na4O(AlH4)2. It is essential to note that the excessive radius of anionic groups inevitably leads to lattice distortion, resulting in asymmetric migration paths and a limited improvement in carrier migration rate. Na4O(AlH4)2 provides a clear example of where Na migrates in two distinct environments. In addition, due to different spatial charge distributions, the interaction strength between anionic groups and Na is different. Strong interactions can cause carriers to appear on a swing, leading to a decrease in conductivity. The low conductivity of Na4O(BF4)2 is a typical example. This study demonstrates that Na4O(BH4)2 exhibits remarkable mechanical and dynamic stability and shows ionic conductivity, two orders of magnitude higher than that of Na4OBr2. This is attributed to the expansion of the carrier migration channels by the anion groups, the moderate interaction between carriers and anionic groups, and the paddle-wheel effect generated by the anion groups, indicating that the paddle-wheel effect is still effective in low-dimensional anti-perovskite structures, in which atoms are arranged asymmetrically.

cond-mat.mtrl-sci↗

arXiv2023

Coordination Engineering in Zirconium-Nitrogen-Functionalized Materials for N2 Reduction: A First-Principles Simulation

Coordination engineering was employed to optimize the coordination environment of the Zr atom anchored on the porphyrins (PP). Five promising ZrPP-A candidates as electrocatalysts for nitrogen reduction reaction (NRR) were identified through a four-step screening strategy. First-principles calculations were utilized to evaluate the performance of the candidate electrocatalysts for NRR. A comprehensive search for reaction pathways revealed that NRR reactions with these selected catalysts tend to follow a hybrid pathway. It is found that orbital hybridization and charge transfer between Zr and its coordination atoms, as well as between ZrPP-A and the adsorbed N2 ensured the stability and high catalytic activity of these selected ZrPP-A. Zr plays a crucial role in coordinating charge transfer during the NRR process. Simultaneously, the coordinating atoms and the PP moiety jointly provide additional charge transfers to or from the adsorbate. An asymmetric coordination environment results in an asymmetric charge distribution of the substrate, causing the adsorbed polarized N2 molecule oriented toward the asymmetric charge aggregation region. Our work underscores the importance of considering not only the single-atom catalyst itself but also its coordination environment for the rational design of efficient catalysts.

cond-mat.mtrl-sci↗

arXiv2023

GUGA-based MRCI approach with Core-Valence Separation Approximation (CVS) for the calculation of the Core-Excited States of molecules

We develop and demonstrate how to use the GUGA-based MRCISD with Core-Valence Separation approximation (CVS) to compute the core-excited states. Firstly, perform a normal SCF or valence MCSCF calculation to optimize the molecular orbitals. Secondly, rotate the optimized target core orbitals and append to the active space, form an extended CVS active space, and perform a CVS-MCSCF calculation for core-excited states. Finally, construct the CVS-MRCI expansion space, and perform a CVS-MRCI calculation to optimize the CI coefficients based on the variational method. The CVS approximation with GUGA-based methods can be implemented by flexible truncation of the Distinct Row Table (DRT). Eliminating the valence-excited configurations from the CVS-MRCI expansion space can prevent variational collapse in the Davidson iteration diagonalization. The accuracy of the CVS-MRCI scheme was investigated for excitation energies and compared with that of the CVS-MCSCF method. The results show that CVS-MRCI is capable of reproducing well-matched vertical core excitation energies that are consistent with experiments, by combining large basis sets and a rational reference space. The calculation results also highlight the fact that the dynamic correlation between electrons makes an undeniable contribution in core-excited states.

physics.comp-ph↗

arXiv2023

Ripple sparse self-attention for monaural speech enhancement

The use of Transformer represents a recent success in speech enhancement. However, as its core component, self-attention suffers from quadratic complexity, which is computationally prohibited for long speech recordings. Moreover, it allows each time frame to attend to all time frames, neglecting the strong local correlations of speech signals. This study presents a simple yet effective sparse self-attention for speech enhancement, called ripple attention, which simultaneously performs fine- and coarse-grained modeling for local and global dependencies, respectively. Specifically, we employ local band attention to enable each frame to attend to its closest neighbor frames in a window at fine granularity, while employing dilated attention outside the window to model the global dependencies at a coarse granularity. We evaluate the efficacy of our ripple attention for speech enhancement on two commonly used training objectives. Extensive experimental results consistently confirm the superior performance of the ripple attention design over standard full self-attention, blockwise attention, and dual-path attention (Sep-Former) in terms of speech quality and intelligibility.

cs.SD↗

arXiv2023

Generative Steganographic Flow

Generative steganography (GS) is a new data hiding manner, featuring direct generation of stego media from secret data. Existing GS methods are generally criticized for their poor performances. In this paper, we propose a novel flow based GS approach -- Generative Steganographic Flow (GSF), which provides direct generation of stego images without cover image. We take the stego image generation and secret data recovery process as an invertible transformation, and build a reversible bijective mapping between input secret data and generated stego images. In the forward mapping, secret data is hidden in the input latent of Glow model to generate stego images. By reversing the mapping, hidden data can be extracted exactly from generated stego images. Furthermore, we propose a novel latent optimization strategy to improve the fidelity of stego images. Experimental results show our proposed GSF has far better performances than SOTA works.

cs.CV↗

arXiv2023

Limits to the strain engineering of layered square-planar nickelate thin films

The layered square-planar nickelates, Nd$_{n+1}$Ni$_{n}$O$_{2n+2}$, are an appealing system to tune the electronic properties of square-planar nickelates via dimensionality; indeed, superconductivity was recently observed in Nd$_{6}$Ni$_{5}$O$_{12}$ thin films. Here, we investigate the role of epitaxial strain in the competing requirements for the synthesis of the $n=3$ Ruddlesden-Popper compound, Nd$_{4}$Ni$_{3}$O$_{10}$, and subsequent reduction to the square-planar phase, Nd$_{4}$Ni$_{3}$O$_{8}$. We synthesize our highest quality Nd$_{4}$Ni$_{3}$O$_{10}$ films under compressive strain on LaAlO$_{3}$ (001), while Nd$_{4}$Ni$_{3}$O$_{10}$ on NdGaO$_{3}$ (110) exhibits tensile strain-induced rock salt faults but retains bulk-like transport properties. A high density of extended defects forms in Nd$_{4}$Ni$_{3}$O$_{10}$ on SrTiO$_{3}$ (001). Films reduced on LaAlO$_{3}$ become insulating and form compressive strain-induced $c$-axis canting defects, while Nd$_{4}$Ni$_{3}$O$_{8}$ films on NdGaO$_{3}$ are metallic. This work provides a pathway to the synthesis of Nd$_{n+1}$Ni$_{n}$O$_{2n+2}$ thin films and sets limits on the ability to strain engineer these compounds via epitaxy.

cond-mat.mtrl-sci↗

arXiv2022

Antiferromagnetic metal phase in an electron-doped rare-earth nickelate

Long viewed as passive elements, antiferromagnetic materials have emerged as promising candidates for spintronic devices due to their insensitivity to external fields and potential for high-speed switching. Recent work exploiting spin and orbital effects has identified ways to electrically control and probe the spins in metallic antiferromagnets, especially in noncollinear or noncentrosymmetric spin structures. The rare earth nickelate NdNiO3 is known to be a noncollinear antiferromagnet where the onset of antiferromagnetic ordering is concomitant with a transition to an insulating state. Here, we find that for low electron doping, the magnetic order on the nickel site is preserved while electronically a new metallic phase is induced. We show that this metallic phase has a Fermi surface that is mostly gapped by an electronic reconstruction driven by the bond disproportionation. Furthermore, we demonstrate the ability to write to and read from the spin structure via a large zero-field planar Hall effect. Our results expand the already rich phase diagram of the rare-earth nickelates and may enable spintronics applications in this family of correlated oxides.

cond-mat.str-el↗

arXiv2022

Fully Attentional Network for Semantic Segmentation

Recent non-local self-attention methods have proven to be effective in capturing long-range dependencies for semantic segmentation. These methods usually form a similarity map of RC*C (by compressing spatial dimensions) or RHW*HW (by compressing channels) to describe the feature relations along either channel or spatial dimensions, where C is the number of channels, H and W are the spatial dimensions of the input feature map. However, such practices tend to condense feature dependencies along the other dimensions,hence causing attention missing, which might lead to inferior results for small/thin categories or inconsistent segmentation inside large objects. To address this problem, we propose anew approach, namely Fully Attentional Network (FLANet),to encode both spatial and channel attentions in a single similarity map while maintaining high computational efficiency. Specifically, for each channel map, our FLANet can harvest feature responses from all other channel maps, and the associated spatial positions as well, through a novel fully attentional module. Our new method has achieved state-of-the-art performance on three challenging semantic segmentation datasets,i.e., 83.6%, 46.99%, and 88.5% on the Cityscapes test set,the ADE20K validation set, and the PASCAL VOC test set,respectively.

cs.CV↗

arXiv2022

Synthesis and electronic properties of Nd$_{n+1}$Ni$_{n}$O$_{3n+1}$ Ruddlesden-Popper nickelate thin films

The rare-earth nickelates possess a diverse set of collective phenomena including metal-to-insulator transitions, magnetic phase transitions, and, upon chemical reduction, superconductivity. Here, we demonstrate epitaxial stabilization of layered nickelates in the Ruddlesden-Popper form, Nd$_{n+1}$Ni$_n$O$_{3n+1}$, using molecular beam epitaxy. By optimizing the stoichiometry of the parent perovskite NdNiO$_3$, we can reproducibly synthesize the $n = 1 - 5$ member compounds. X-ray absorption spectroscopy at the O $K$ and Ni $L$ edges indicate systematic changes in both the nickel-oxygen hybridization level and nominal nickel filling from 3$d^8$ to 3$d^7$ as we move across the series from $n = 1$ to $n = \infty$. The $n = 3 - 5$ compounds exhibit weakly hysteretic metal-to-insulator transitions with transition temperatures that depress with increasing order toward NdNiO$_3$ ($n = \infty)$.

cond-mat.mtrl-sci↗

arXiv2022

Stochastic Planner-Actor-Critic for Unsupervised Deformable Image Registration

Large deformations of organs, caused by diverse shapes and nonlinear shape changes, pose a significant challenge for medical image registration. Traditional registration methods need to iteratively optimize an objective function via a specific deformation model along with meticulous parameter tuning, but which have limited capabilities in registering images with large deformations. While deep learning-based methods can learn the complex mapping from input images to their respective deformation field, it is regression-based and is prone to be stuck at local minima, particularly when large deformations are involved. To this end, we present Stochastic Planner-Actor-Critic (SPAC), a novel reinforcement learning-based framework that performs step-wise registration. The key notion is warping a moving image successively by each time step to finally align to a fixed image. Considering that it is challenging to handle high dimensional continuous action and state spaces in the conventional reinforcement learning (RL) framework, we introduce a new concept `Plan' to the standard Actor-Critic model, which is of low dimension and can facilitate the actor to generate a tractable high dimensional action. The entire framework is based on unsupervised training and operates in an end-to-end manner. We evaluate our method on several 2D and 3D medical image datasets, some of which contain large deformations. Our empirical results highlight that our work achieves consistent, significant gains and outperforms state-of-the-art methods.

eess.IV↗

arXiv2022

Time-Frequency Attention for Monaural Speech Enhancement

Most studies on speech enhancement generally don't consider the energy distribution of speech in time-frequency (T-F) representation, which is important for accurate prediction of mask or spectra. In this paper, we present a simple yet effective T-F attention (TFA) module, where a 2-D attention map is produced to provide differentiated weights to the spectral components of T-F representation. To validate the effectiveness of our proposed TFA module, we use the residual temporal convolution network (ResTCN) as the backbone network and conduct extensive experiments on two commonly used training targets. Our experiments demonstrate that applying our TFA module significantly improves the performance in terms of five objective evaluation metrics with negligible parameter overhead. The evaluation results show that the proposed ResTCN with the TFA module (ResTCN+TFA) consistently outperforms other baselines by a large margin.

cs.SD↗

arXiv2022

Synergistic Network Learning and Label Correction for Noise-robust Image Classification

Large training datasets almost always contain examples with inaccurate or incorrect labels. Deep Neural Networks (DNNs) tend to overfit training label noise, resulting in poorer model performance in practice. To address this problem, we propose a robust label correction framework combining the ideas of small loss selection and noise correction, which learns network parameters and reassigns ground truth labels iteratively. Taking the expertise of DNNs to learn meaningful patterns before fitting noise, our framework first trains two networks over the current dataset with small loss selection. Based on the classification loss and agreement loss of two networks, we can measure the confidence of training data. More and more confident samples are selected for label correction during the learning process. We demonstrate our method on both synthetic and real-world datasets with different noise types and rates, including CIFAR-10, CIFAR-100 and Clothing1M, where our method outperforms the baseline approaches.

cs.CV↗

arXiv2022

Recursive Least Squares Policy Control with Echo State Network

The echo state network (ESN) is a special type of recurrent neural networks for processing the time-series dataset. However, limited by the strong correlation among sequential samples of the agent, ESN-based policy control algorithms are difficult to use the recursive least squares (RLS) algorithm to update the ESN's parameters. To solve this problem, we propose two novel policy control algorithms, ESNRLS-Q and ESNRLS-Sarsa. Firstly, to reduce the correlation of training samples, we use the leaky integrator ESN and the mini-batch learning mode. Secondly, to make RLS suitable for training ESN in mini-batch mode, we present a new mean-approximation method for updating the RLS correlation matrix. Thirdly, to prevent ESN from over-fitting, we use the L1 regularization technique. Lastly, to prevent the target state-action value from overestimation, we employ the Mellowmax method. Simulation results show that our algorithms have good convergence performance.

cs.LG↗

arXiv2022

Recursive Least Squares for Training and Pruning Convolutional Neural Networks

Convolutional neural networks (CNNs) have succeeded in many practical applications. However, their high computation and storage requirements often make them difficult to deploy on resource-constrained devices. In order to tackle this issue, many pruning algorithms have been proposed for CNNs, but most of them can't prune CNNs to a reasonable level. In this paper, we propose a novel algorithm for training and pruning CNNs based on the recursive least squares (RLS) optimization. After training a CNN for some epochs, our algorithm combines inverse input autocorrelation matrices and weight matrices to evaluate and prune unimportant input channels or nodes layer by layer. Then, our algorithm will continue to train the pruned network, and won't do the next pruning until the pruned network recovers the full performance of the old network. Besides for CNNs, the proposed algorithm can be used for feedforward neural networks (FNNs). Three experiments on MNIST, CIFAR-10 and SVHN datasets show that our algorithm can achieve the more reasonable pruning and have higher learning efficiency than other four popular pruning algorithms.

cs.LG↗

arXiv2021

Stochastic Actor-Executor-Critic for Image-to-Image Translation

Training a model-free deep reinforcement learning model to solve image-to-image translation is difficult since it involves high-dimensional continuous state and action spaces. In this paper, we draw inspiration from the recent success of the maximum entropy reinforcement learning framework designed for challenging continuous control problems to develop stochastic policies over high dimensional continuous spaces including image representation, generation, and control simultaneously. Central to this method is the Stochastic Actor-Executor-Critic (SAEC) which is an off-policy actor-critic model with an additional executor to generate realistic images. Specifically, the actor focuses on the high-level representation and control policy by a stochastic latent action, as well as explicitly directs the executor to generate low-level actions to manipulate the state. Experiments on several image-to-image translation tasks have demonstrated the effectiveness and robustness of the proposed SAEC when facing high-dimensional continuous space problems.

cs.CV↗

arXiv2021

Denoised Non-Local Neural Network for Semantic Segmentation

The non-local network has become a widely used technique for semantic segmentation, which computes an attention map to measure the relationships of each pixel pair. However, most of the current popular non-local models tend to ignore the phenomenon that the calculated attention map appears to be very noisy, containing inter-class and intra-class inconsistencies, which lowers the accuracy and reliability of the non-local methods. In this paper, we figuratively denote these inconsistencies as attention noises and explore the solutions to denoise them. Specifically, we inventively propose a Denoised Non-Local Network (Denoised NL), which consists of two primary modules, i.e., the Global Rectifying (GR) block and the Local Retention (LR) block, to eliminate the inter-class and intra-class noises respectively. First, GR adopts the class-level predictions to capture a binary map to distinguish whether the selected two pixels belong to the same category. Second, LR captures the ignored local dependencies and further uses them to rectify the unwanted hollows in the attention map. The experimental results on two challenging semantic segmentation datasets demonstrate the superior performance of our model. Without any external training data, our proposed Denoised NL can achieve the state-of-the-art performance of 83.5\% and 46.69\% mIoU on Cityscapes and ADE20K, respectively.

cs.CV↗