arXiv subjects

Hao Zhang

Publications and source records attributed to Hao Zhang.

At least 955 records · Page 53Linked to original sources

arXiv2022

Alternating Synthetic and Real Gradients for Neural Language Modeling

Training recurrent neural networks (RNNs) with backpropagation through time (BPTT) has known drawbacks such as being difficult to capture longterm dependencies in sequences. Successful alternatives to BPTT have not yet been discovered. Recently, BP with synthetic gradients by a decoupled neural interface module has been proposed to replace BPTT for training RNNs. On the other hand, it has been shown that the representations learned with synthetic and real gradients are different though they are functionally identical. In this project, we explore ways of combining synthetic and real gradients with application to neural language modeling tasks. Empirically, we demonstrate the effectiveness of alternating training with synthetic and real gradients after periodic warm restarts on language modeling tasks.

cs.LG↗

arXiv2022

Neural Dual Contouring

We introduce neural dual contouring (NDC), a new data-driven approach to mesh reconstruction based on dual contouring (DC). Like traditional DC, it produces exactly one vertex per grid cell and one quad for each grid edge intersection, a natural and efficient structure for reproducing sharp features. However, rather than computing vertex locations and edge crossings with hand-crafted functions that depend directly on difficult-to-obtain surface gradients, NDC uses a neural network to predict them. As a result, NDC can be trained to produce meshes from signed or unsigned distance fields, binary voxel grids, or point clouds (with or without normals); and it can produce open surfaces in cases where the input represents a sheet or partial surface. During experiments with five prominent datasets, we find that NDC, when trained on one of the datasets, generalizes well to the others. Furthermore, NDC provides better surface reconstruction accuracy, feature preservation, output complexity, triangle quality, and inference time in comparison to previous learned (e.g., neural marching cubes, convolutional occupancy networks) and traditional (e.g., Poisson) methods. Code and data are available at https://github.com/czq142857/NDC.

cs.CV↗

arXiv2022

Why Adversarial Training of ReLU Networks Is Difficult?

This paper mathematically derives an analytic solution of the adversarial perturbation on a ReLU network, and theoretically explains the difficulty of adversarial training. Specifically, we formulate the dynamics of the adversarial perturbation generated by the multi-step attack, which shows that the adversarial perturbation tends to strengthen eigenvectors corresponding to a few top-ranked eigenvalues of the Hessian matrix of the loss w.r.t. the input. We also prove that adversarial training tends to strengthen the influence of unconfident input samples with large gradient norms in an exponential manner. Besides, we find that adversarial training strengthens the influence of the Hessian matrix of the loss w.r.t. network parameters, which makes the adversarial training more likely to oscillate along directions of a few samples, and boosts the difficulty of adversarial training. Crucially, our proofs provide a unified explanation for previous findings in understanding adversarial training.

cs.LG↗

arXiv2022

GALOIS: Boosting Deep Reinforcement Learning via Generalizable Logic Synthesis

Despite achieving superior performance in human-level control problems, unlike humans, deep reinforcement learning (DRL) lacks high-order intelligence (e.g., logic deduction and reuse), thus it behaves ineffectively than humans regarding learning and generalization in complex problems. Previous works attempt to directly synthesize a white-box logic program as the DRL policy, manifesting logic-driven behaviors. However, most synthesis methods are built on imperative or declarative programming, and each has a distinct limitation, respectively. The former ignores the cause-effect logic during synthesis, resulting in low generalizability across tasks. The latter is strictly proof-based, thus failing to synthesize programs with complex hierarchical logic. In this paper, we combine the above two paradigms together and propose a novel Generalizable Logic Synthesis (GALOIS) framework to synthesize hierarchical and strict cause-effect logic programs. GALOIS leverages the program sketch and defines a new sketch-based hybrid program language for guiding the synthesis. Based on that, GALOIS proposes a sketch-based program synthesis method to automatically generate white-box programs with generalizable and interpretable cause-effect logic. Extensive evaluations on various decision-making tasks with complex logic demonstrate the superiority of GALOIS over mainstream baselines regarding the asymptotic performance, generalizability, and great knowledge reusability across different environments.

cs.AI↗

arXiv2022

Active Domain Adaptation with Multi-level Contrastive Units for Semantic Segmentation

To further reduce the cost of semi-supervised domain adaptation (SSDA) labeling, a more effective way is to use active learning (AL) to annotate a selected subset with specific properties. However, domain adaptation tasks are always addressed in two interactive aspects: domain transfer and the enhancement of discrimination, which requires the selected data to be both uncertain under the model and diverse in feature space. Contrary to active learning in classification tasks, it is usually challenging to select pixels that contain both the above properties in segmentation tasks, leading to the complex design of pixel selection strategy. To address such an issue, we propose a novel Active Domain Adaptation scheme with Multi-level Contrastive Units (ADA-MCU) for semantic image segmentation. A simple pixel selection strategy followed with the construction of multi-level contrastive units is introduced to optimize the model for both domain adaptation and active supervised learning. In practice, MCUs are constructed from intra-image, cross-image, and cross-domain levels by using both labeled and unlabeled pixels. At each level, we define contrastive losses from center-to-center and pixel-to-pixel manners, with the aim of jointly aligning the category centers and reducing outliers near the decision boundaries. In addition, we also introduce a categories correlation matrix to implicitly describe the relationship between categories, which are used to adjust the weights of the losses for MCUs. Extensive experimental results on standard benchmarks show that the proposed method achieves competitive performance against state-of-the-art SSDA methods with 50% fewer labeled pixels and significantly outperforms state-of-the-art with a large margin by using the same level of annotation cost.

cs.CV↗

arXiv2022

Understanding temperature-dependent SU($3$) spin dynamics in the $S=1$ antiferromagnet Ba$_2$FeSi$_2$O$_7$

Quantum magnets admit more than one classical limit and $N$-level systems with strong single-ion anisotropy are expected to be described by a classical approximation based on SU($N$) coherent states. Here we test this hypothesis by modeling finite temperature inelastic neutron scattering (INS) data of the effective spin-one antiferromagnet \bfso{}. The measured dynamic structure factor is calculated with a generalized Landau-Lifshitz dynamics for SU($3$) spins. Unlike the traditional classical limit based on SU($2$) coherent states, the results obtained with classical SU($3$) spins are in good agreement with the measured temperature-dependent spectrum. The SU($3$) approach developed here provides a general framework to understand the broad class of materials comprising weakly coupled antiferromagnetic dimers, trimers, or tetramers, and magnets with strong single-ion anisotropy.

cond-mat.str-el↗

arXiv2022

Distantly-Supervised Long-Tailed Relation Extraction Using Constraint Graphs

Label noise and long-tailed distributions are two major challenges in distantly supervised relation extraction. Recent studies have shown great progress on denoising, but paid little attention to the problem of long-tailed relations. In this paper, we introduce a constraint graph to model the dependencies between relation labels. On top of that, we further propose a novel constraint graph-based relation extraction framework(CGRE) to handle the two challenges simultaneously. CGRE employs graph convolution networks to propagate information from data-rich relation nodes to data-poor relation nodes, and thus boosts the representation learning of long-tailed relations. To further improve the noise immunity, a constraint-aware attention module is designed in CGRE to integrate the constraint information. Extensive experimental results indicate that CGRE achieves significant improvements over the previous methods for both denoising and long-tailed relation extraction. The pre-processed datasets and source code are publicly available at https://github.com/tmliang/CGRE.

cs.CL↗

arXiv2022

Downstream Transformer Generation of Question-Answer Pairs with Preprocessing and Postprocessing Pipelines

We present a system called TP3 to perform a downstream task of transformers on generating question-answer pairs (QAPs) from a given article. TP3 first finetunes pretrained transformers on QAP datasets, then uses a preprocessing pipeline to select appropriate answers, feeds the relevant sentences and the answer to the finetuned transformer to generate candidate QAPs, and finally uses a postprocessing pipeline to filter inadequate QAPs. In particular, using pretrained T5 models as transformers and the SQuAD dataset as the finetruning dataset, we show that TP3 generates satisfactory number of QAPs with high qualities on the Gaokao-EN dataset.

cs.CL↗

arXiv2022

Manifoldron: Direct Space Partition via Manifold Discovery

A neural network with the widely-used ReLU activation has been shown to partition the sample space into many convex polytopes for prediction. However, the parameterized way a neural network and other machine learning models use to partition the space has imperfections, \textit{e}.\textit{g}., the compromised interpretability for complex models, the inflexibility in decision boundary construction due to the generic character of the model, and the risk of being trapped into shortcut solutions. In contrast, although the non-parameterized models can adorably avoid or downplay these issues, they are usually insufficiently powerful either due to over-simplification or the failure to accommodate the manifold structures of data. In this context, we first propose a new type of machine learning models referred to as Manifoldron that directly derives decision boundaries from data and partitions the space via manifold structure discovery. Then, we systematically analyze the key characteristics of the Manifoldron such as manifold characterization capability and its link to neural networks. The experimental results on 4 synthetic examples, 20 public benchmark datasets, and 1 real-world application demonstrate that the proposed Manifoldron performs competitively compared to the mainstream machine learning models. We have shared our code in \url{https://github.com/wdayang/Manifoldron} for free download and evaluation.

cs.LG↗

arXiv2022

FedCos: A Scene-adaptive Federated Optimization Enhancement for Performance Improvement

As an emerging technology, federated learning (FL) involves training machine learning models over distributed edge devices, which attracts sustained attention and has been extensively studied. However, the heterogeneity of client data severely degrades the performance of FL compared with that in centralized training. It causes the locally trained models of clients to move in different directions. On the one hand, it slows down or even stalls the global updates, leading to inefficient communication. On the other hand, it enlarges the distances between local models, resulting in an aggregated global model with poor performance. Fortunately, these shortcomings can be mitigated by reducing the angle between the directions that local models move in. Based on this fact, we propose FedCos, which reduces the directional inconsistency of local models by introducing a cosine-similarity penalty. It promotes the local model iterations towards an auxiliary global direction. Moreover, our approach is auto-adapt to various non-IID settings without an elaborate selection of hyperparameters. The experimental results show that FedCos outperforms the well-known baselines and can enhance them under a variety of FL scenes, including varying degrees of data heterogeneity, different number of participants, and cross-silo and cross-device settings. Besides, FedCos improves communication efficiency by 2 to 5 times. With the help of FedCos, multiple FL methods require significantly fewer communication rounds than before to obtain a model with comparable performance.

cs.LG↗

arXiv2022

Direct visualization of ultrafast lattice ordering triggered by an electron-hole plasma in 2D perovskites

Direct visualization of ultrafast coupling between charge carriers and lattice degrees of freedom in photo-excited semiconductors has remained a long-standing challenge and is critical for understanding the light-induced physical behavior of materials under extreme non-equilibrium conditions. Here, by monitoring the evolution of the wave-vector resolved ultrafast electron diffraction intensity following above-bandgap photo-excitation, we obtain a direct visual of the structural dynamics in monocrystalline 2D perovskites. Analysis reveals a surprising, light-induced ultrafast lattice ordering resulting from a strong interaction between hot-carriers and the perovskite lattice, which induces an in-plane octahedra rotation, towards a more symmetric phase. Correlated ultrafast spectroscopy performed at the same carrier density as ultrafast electron diffraction reveals that the creation of a hot and dense electron-hole plasma triggers lattice ordering at short timescales by modulating the crystal cohesive energy. Finally, we show that the interaction between the carrier gas and the lattice can be altered by tailoring the rigidity of the 2D perovskite by choosing the appropriate organic spacer layer.

cond-mat.mtrl-sci↗

arXiv2022

Exceptional Point modulated by Kerr effect in Anti-Parity-Time Symmetry System

With respect to parity-time (PT) symmetry, anti-parity-time (APT) symmetric system exhibits much easier readout mechanism due to its real frequency splitting. Generally, such systems need to be operated at exceptional points (EPs) to obtain the best performance. However, strict conditons to locate APT symmetric systems at their EPs precisely put restraints on their practical applications. To overcome this problem, we propose a scheme to manipulate the EPs in APT symmetric configuration by Kerr effect. It is demonstrated that operating EPs by self-phase modulation alone will impede the frequency splitting caused by external perturbations, while cross-phase modulation can enhance the response to measurable perturbations. We also investigate the thermal effect induced by high light intensity, which could reduce the power to manipulate EPs. This proposed scheme can pave a new way in fabricating devices based on APT symmetry.

physics.optics↗

arXiv2022

UNIST: Unpaired Neural Implicit Shape Translation Network

We introduce UNIST, the first deep neural implicit model for general-purpose, unpaired shape-to-shape translation, in both 2D and 3D domains. Our model is built on autoencoding implicit fields, rather than point clouds which represents the state of the art. Furthermore, our translation network is trained to perform the task over a latent grid representation which combines the merits of both latent-space processing and position awareness, to not only enable drastic shape transforms but also well preserve spatial features and fine local details for natural shape translations. With the same network architecture and only dictated by the input domain pairs, our model can learn both style-preserving content alteration and content-preserving style transfer. We demonstrate the generality and quality of the translation results, and compare them to well-known baselines. Code is available at https://qiminchen.github.io/unist/.

cs.CV↗

arXiv2022

DAB-DETR: Dynamic Anchor Boxes are Better Queries for DETR

We present in this paper a novel query formulation using dynamic anchor boxes for DETR (DEtection TRansformer) and offer a deeper understanding of the role of queries in DETR. This new formulation directly uses box coordinates as queries in Transformer decoders and dynamically updates them layer-by-layer. Using box coordinates not only helps using explicit positional priors to improve the query-to-feature similarity and eliminate the slow training convergence issue in DETR, but also allows us to modulate the positional attention map using the box width and height information. Such a design makes it clear that queries in DETR can be implemented as performing soft ROI pooling layer-by-layer in a cascade manner. As a result, it leads to the best performance on MS-COCO benchmark among the DETR-like detection models under the same setting, e.g., AP 45.7\% using ResNet50-DC5 as backbone trained in 50 epochs. We also conducted extensive experiments to confirm our analysis and verify the effectiveness of our methods. Code is available at \url{https://github.com/SlongLiu/DAB-DETR}.

cs.CV↗

arXiv2022

CP$^2$ Skyrmions and Skyrmion Crystals in Realistic Quantum Magnets

Magnetic skyrmions are nanoscale topological textures that have been recently observed in different families of quantum magnets. These textures are known as CP$^1$ skyrmions because the target manifold of the magnetization field is the 2D sphere $S^2 \cong$ CP$^{1}$. Here we report the emergence of magnetic CP$^2$ skyrmions in realistic spin-$1$ models. Unlike CP$^1$ skyrmions, CP$^2$ skyrmions can also arise as metastable textures of quantum paramagnets, opening a new road to discover emergent topological solitons in non-magnetic materials. The quantum phase diagram of the spin-$1$ models also includes magnetic field-induced CP$^2$ skyrmion crystals that can be detected with regular momentum- (diffraction) and real-space (Lorentz transmission electron microscopy) experimental techniques.

cond-mat.str-el↗

arXiv2022

RIM-Net: Recursive Implicit Fields for Unsupervised Learning of Hierarchical Shape Structures

We introduce RIM-Net, a neural network which learns recursive implicit fields for unsupervised inference of hierarchical shape structures. Our network recursively decomposes an input 3D shape into two parts, resulting in a binary tree hierarchy. Each level of the tree corresponds to an assembly of shape parts, represented as implicit functions, to reconstruct the input shape. At each node of the tree, simultaneous feature decoding and shape decomposition are carried out by their respective feature and part decoders, with weight sharing across the same hierarchy level. As an implicit field decoder, the part decoder is designed to decompose a sub-shape, via a two-way branched reconstruction, where each branch predicts a set of parameters defining a Gaussian to serve as a local point distribution for shape reconstruction. With reconstruction losses accounted for at each hierarchy level and a decomposition loss at each node, our network training does not require any ground-truth segmentations, let alone hierarchies. Through extensive experiments and comparisons to state-of-the-art alternatives, we demonstrate the quality, consistency, and interpretability of hierarchical structural inference by RIM-Net.

cs.CV↗

arXiv2022

Group Contextualization for Video Recognition

Learning discriminative representation from the complex spatio-temporal dynamic space is essential for video recognition. On top of those stylized spatio-temporal computational units, further refining the learnt feature with axial contexts is demonstrated to be promising in achieving this goal. However, previous works generally focus on utilizing a single kind of contexts to calibrate entire feature channels and could hardly apply to deal with diverse video activities. The problem can be tackled by using pair-wise spatio-temporal attentions to recompute feature response with cross-axis contexts at the expense of heavy computations. In this paper, we propose an efficient feature refinement method that decomposes the feature channels into several groups and separately refines them with different axial contexts in parallel. We refer this lightweight feature calibration as group contextualization (GC). Specifically, we design a family of efficient element-wise calibrators, i.e., ECal-G/S/T/L, where their axial contexts are information dynamics aggregated from other axes either globally or locally, to contextualize feature channel groups. The GC module can be densely plugged into each residual layer of the off-the-shelf video networks. With little computational overhead, consistent improvement is observed when plugging in GC on different networks. By utilizing calibrators to embed feature with four different kinds of contexts in parallel, the learnt representation is expected to be more resilient to diverse types of activities. On videos with rich temporal variations, empirically GC can boost the performance of 2D-CNN (e.g., TSN and TSM) to a level comparable to the state-of-the-art video networks. Code is available at https://github.com/haoyanbin918/Group-Contextualization.

cs.CV↗

arXiv2022

Quantitative Performance Assessment of CNN Units via Topological Entropy Calculation

Identifying the status of individual network units is critical for understanding the mechanism of convolutional neural networks (CNNs). However, it is still challenging to reliably give a general indication of unit status, especially for units in different network models. To this end, we propose a novel method for quantitatively clarifying the status of single unit in CNN using algebraic topological tools. Unit status is indicated via the calculation of a defined topological-based entropy, called feature entropy, which measures the degree of chaos of the global spatial pattern hidden in the unit for a category. In this way, feature entropy could provide an accurate indication of status for units in different networks with diverse situations like weight-rescaling operation. Further, we show that feature entropy decreases as the layer goes deeper and shares almost simultaneous trend with loss during training. We show that by investigating the feature entropy of units on only training data, it could give discrimination between networks with different generalization ability from the view of the effectiveness of feature representations.

cs.CV↗