arXiv subjects

Hong Liu

Publications and source records attributed to Hong Liu.

At least 217 records · Page 12Linked to original sources

arXiv2024

Hourglass Tokenizer for Efficient Transformer-Based 3D Human Pose Estimation

Transformers have been successfully applied in the field of video-based 3D human pose estimation. However, the high computational costs of these video pose transformers (VPTs) make them impractical on resource-constrained devices. In this paper, we present a plug-and-play pruning-and-recovering framework, called Hourglass Tokenizer (HoT), for efficient transformer-based 3D human pose estimation from videos. Our HoT begins with pruning pose tokens of redundant frames and ends with recovering full-length tokens, resulting in a few pose tokens in the intermediate transformer blocks and thus improving the model efficiency. To effectively achieve this, we propose a token pruning cluster (TPC) that dynamically selects a few representative tokens with high semantic diversity while eliminating the redundancy of video frames. In addition, we develop a token recovering attention (TRA) to restore the detailed spatio-temporal information based on the selected tokens, thereby expanding the network output to the original full-length temporal resolution for fast inference. Extensive experiments on two benchmark datasets (i.e., Human3.6M and MPI-INF-3DHP) demonstrate that our method can achieve both high efficiency and estimation accuracy compared to the original VPT models. For instance, applying to MotionBERT and MixSTE on Human3.6M, our HoT can save nearly 50% FLOPs without sacrificing accuracy and nearly 40% FLOPs with only 0.2% accuracy drop, respectively. Code and models are available at https://github.com/NationalGAILab/HoT.

cs.CV↗

arXiv2024

Identity-aware Dual-constraint Network for Cloth-Changing Person Re-identification

Cloth-Changing Person Re-Identification (CC-ReID) aims to accurately identify the target person in more realistic surveillance scenarios, where pedestrians usually change their clothing. Despite great progress, limited cloth-changing training samples in existing CC-ReID datasets still prevent the model from adequately learning cloth-irrelevant features. In addition, due to the absence of explicit supervision to keep the model constantly focused on cloth-irrelevant areas, existing methods are still hampered by the disruption of clothing variations. To solve the above issues, we propose an Identity-aware Dual-constraint Network (IDNet) for the CC-ReID task. Specifically, to help the model extract cloth-irrelevant clues, we propose a Clothes Diversity Augmentation (CDA), which generates more realistic cloth-changing samples by enriching the clothing color while preserving the texture. In addition, a Multi-scale Constraint Block (MCB) is designed, which extracts fine-grained identity-related features and effectively transfers cloth-irrelevant knowledge. Moreover, a Counterfactual-guided Attention Module (CAM) is presented, which learns cloth-irrelevant features from channel and space dimensions and utilizes the counterfactual intervention for supervising the attention map to highlight identity-related regions. Finally, a Semantic Alignment Constraint (SAC) is designed to facilitate high-level semantic feature interaction. Comprehensive experiments on four CC-ReID datasets indicate that our method outperforms prior state-of-the-art approaches.

cs.CV↗

arXiv2024

Feature Completion Transformer for Occluded Person Re-identification

Occluded person re-identification (Re-ID) is a challenging problem due to the destruction of occluders. Most existing methods focus on visible human body parts through some prior information. However, when complementary occlusions occur, features in occluded regions can interfere with matching, which affects performance severely. In this paper, different from most previous works that discard the occluded region, we propose a Feature Completion Transformer (FCFormer) to implicitly complement the semantic information of occluded parts in the feature space. Specifically, Occlusion Instance Augmentation (OIA) is proposed to simulates real and diverse occlusion situations on the holistic image. These augmented images not only enrich the amount of occlusion samples in the training set, but also form pairs with the holistic images. Subsequently, a dual-stream architecture with a shared encoder is proposed to learn paired discriminative features from pairs of inputs. Without additional semantic information, an occluded-holistic feature sample-label pair can be automatically created. Then, Feature Completion Decoder (FCD) is designed to complement the features of occluded regions by using learnable tokens to aggregate possible information from self-generated occluded features. Finally, we propose the Cross Hard Triplet (CHT) loss to further bridge the gap between complementing features and extracting features under the same ID. In addition, Feature Completion Consistency (FC$^2$) loss is introduced to help the generated completion feature distribution to be closer to the real holistic feature distribution. Extensive experiments over five challenging datasets demonstrate that the proposed FCFormer achieves superior performance and outperforms the state-of-the-art methods by significant margins on occluded datasets.

cs.CV↗

arXiv2024

Generalized Ramsey--Turán density for cliques

We study the generalized Ramsey--Turán function $\mathrm{RT}(n,K_s,K_t,o(n))$, which is the maximum possible number of copies of $K_s$ in an $n$-vertex $K_t$-free graph with independence number $o(n)$. The case when $s=2$ was settled by Erd{ő}s, S{ó}s, Bollob{á}s, Hajnal, and Szemerédi in the 1980s. We combinatorially resolve the general case for all $s\ge 3$, showing that the (asymptotic) extremal graphs for this problem have simple (bounded) structures. In particular, it implies that the extremal structures follow a periodic pattern when $t$ is much larger than $s$. Our results disprove a conjecture of Balogh, Liu, and Sharifzadeh and show that a relaxed version does hold.

math.CO↗

arXiv2024

Federated Modality-specific Encoders and Multimodal Anchors for Personalized Brain Tumor Segmentation

Most existing federated learning (FL) methods for medical image analysis only considered intramodal heterogeneity, limiting their applicability to multimodal imaging applications. In practice, it is not uncommon that some FL participants only possess a subset of the complete imaging modalities, posing inter-modal heterogeneity as a challenge to effectively training a global model on all participants' data. In addition, each participant would expect to obtain a personalized model tailored for its local data characteristics from the FL in such a scenario. In this work, we propose a new FL framework with federated modality-specific encoders and multimodal anchors (FedMEMA) to simultaneously address the two concurrent issues. Above all, FedMEMA employs an exclusive encoder for each modality to account for the inter-modal heterogeneity in the first place. In the meantime, while the encoders are shared by the participants, the decoders are personalized to meet individual needs. Specifically, a server with full-modal data employs a fusion decoder to aggregate and fuse representations from all modality-specific encoders, thus bridging the modalities to optimize the encoders via backpropagation reversely. Meanwhile, multiple anchors are extracted from the fused multimodal representations and distributed to the clients in addition to the encoder parameters. On the other end, the clients with incomplete modalities calibrate their missing-modal representations toward the global full-modal anchors via scaled dot-product cross-attention, making up the information loss due to absent modalities while adapting the representations of present ones. FedMEMA is validated on the BraTS 2020 benchmark for multimodal brain tumor segmentation. Results show that it outperforms various up-to-date methods for multimodal and personalized FL and that its novel designs are effective. Our code is available.

cs.CV↗

arXiv2024

WSI-SAM: Multi-resolution Segment Anything Model (SAM) for histopathology whole-slide images

The Segment Anything Model (SAM) marks a significant advancement in segmentation models, offering robust zero-shot abilities and dynamic prompting. However, existing medical SAMs are not suitable for the multi-scale nature of whole-slide images (WSIs), restricting their effectiveness. To resolve this drawback, we present WSI-SAM, enhancing SAM with precise object segmentation capabilities for histopathology images using multi-resolution patches, while preserving its efficient, prompt-driven design, and zero-shot abilities. To fully exploit pretrained knowledge while minimizing training overhead, we keep SAM frozen, introducing only minimal extra parameters and computational overhead. In particular, we introduce High-Resolution (HR) token, Low-Resolution (LR) token and dual mask decoder. This decoder integrates the original SAM mask decoder with a lightweight fusion module that integrates features at multiple scales. Instead of predicting a mask independently, we integrate HR and LR token at intermediate layer to jointly learn features of the same object across multiple resolutions. Experiments show that our WSI-SAM outperforms state-of-the-art SAM and its variants. In particular, our model outperforms SAM by 4.1 and 2.5 percent points on a ductal carcinoma in situ (DCIS) segmentation tasks and breast cancer metastasis segmentation task (CAMELYON16 dataset). The code will be available at https://github.com/HongLiuuuuu/WSI-SAM.

cs.CV↗

arXiv2024

Sophia: A Scalable Stochastic Second-order Optimizer for Language Model Pre-training

Given the massive cost of language model pre-training, a non-trivial improvement of the optimization algorithm would lead to a material reduction on the time and cost of training. Adam and its variants have been state-of-the-art for years, and more sophisticated second-order (Hessian-based) optimizers often incur too much per-step overhead. In this paper, we propose Sophia, Second-order Clipped Stochastic Optimization, a simple scalable second-order optimizer that uses a light-weight estimate of the diagonal Hessian as the pre-conditioner. The update is the moving average of the gradients divided by the moving average of the estimated Hessian, followed by element-wise clipping. The clipping controls the worst-case update size and tames the negative impact of non-convexity and rapid change of Hessian along the trajectory. Sophia only estimates the diagonal Hessian every handful of iterations, which has negligible average per-step time and memory overhead. On language modeling with GPT models of sizes ranging from 125M to 1.5B, Sophia achieves a 2x speed-up compared to Adam in the number of steps, total compute, and wall-clock time, achieving the same perplexity with 50% fewer steps, less total compute, and reduced wall-clock time. Theoretically, we show that Sophia, in a much simplified setting, adapts to the heterogeneous curvatures in different parameter dimensions, and thus has a run-time bound that does not depend on the condition number of the loss.

cs.LG↗

arXiv2024

GraphTranslator: Aligning Graph Model to Large Language Model for Open-ended Tasks

Large language models (LLMs) like ChatGPT, exhibit powerful zero-shot and instruction-following capabilities, have catalyzed a revolutionary transformation across diverse fields, especially for open-ended tasks. While the idea is less explored in the graph domain, despite the availability of numerous powerful graph models (GMs), they are restricted to tasks in a pre-defined form. Although several methods applying LLMs to graphs have been proposed, they fail to simultaneously handle the pre-defined and open-ended tasks, with LLM as a node feature enhancer or as a standalone predictor. To break this dilemma, we propose to bridge the pretrained GM and LLM by a Translator, named GraphTranslator, aiming to leverage GM to handle the pre-defined tasks effectively and utilize the extended interface of LLMs to offer various open-ended tasks for GM. To train such Translator, we propose a Producer capable of constructing the graph-text alignment data along node information, neighbor information and model information. By translating node representation into tokens, GraphTranslator empowers an LLM to make predictions based on language instructions, providing a unified perspective for both pre-defined and open-ended tasks. Extensive results demonstrate the effectiveness of our proposed GraphTranslator on zero-shot node classification. The graph question answering experiments reveal our GraphTranslator potential across a broad spectrum of open-ended tasks through language instructions. Our code is available at: https://github.com/alibaba/GraphTranslator.

cs.AI↗

arXiv2024

Graph with any rational density and no rich subsets of linear size

A well-known application of the dependent random choice asserts that any $n$-vertex graph $G$ with positive edge density contains a `rich' vertex subset $U$ of size $n^{1-o(1)}$ such that every pair of vertices in $U$ has at least $n^{1-o(1)}$ common neighbors. In 2003, using a beautiful construction on hypercube, Kostochka and Sudakov showed that this is tight: one cannot remove the $o(1)$ terms even if the edge density of $G$ is $1/2$. In this paper, we generalize their result from pairs to tuples. To be precise, we show that given every pair of positive integers $p<q$, there is an $n$-vertex graph $G$ for all sufficiently large $n$ with edge density $p/q$ such that any vertex subset $U$ of size $Ω(n)$ contains $q$ vertices, any $p+1$ of which have $o(n)$ common neighbors. The edge density $p/q$ is best possible. Our construction uses isoperimetry and concentration of measure on high dimensional complex spheres.

math.CO↗

arXiv2024

The photodissociation dynamics and ultrafast electron diffraction image of cyclobutanone from the surface hopping dynamics simulation

The comprehension of nonadiabatic dynamics in polyatomic systems relies heavily on the simultaneous advancements in theoretical and experimental domains. The gas-phase electron diffraction (GUED) technique has attracted widespread attention as a promising tool for observing the photochemical and photophysical features at all-atomic level with high temporal and spatial resolutions. In this work, the GUED spectra were predicted to perform a double-blind test of accuracy in excited-state simulation for cyclobutanone based on the trajectory surface hopping method, with respect to the benchmark data obtained by upcoming MeV-GUED experiments at the Stanfold Linear Accelerator Laboratory. The results show that the ultrafast nonadiabatic dynamics occurs in the photoinduced dynamics, and two C2 and C3 channels play dominant roles in the nonadiabatic reactions of cyclobutanone. The simulated UED signal can be directly interpreted by atomic movements, providing a unique view to monitor the time-dependent evolution of the molecular structure in the femtosecond dynamics.

physics.chem-ph↗

arXiv2024

How connectivity affects the extremal number of trees

The Erdős-Sós conjecture states that the maximum number of edges in an $n$-vertex graph without a given $k$-vertex tree is at most $\frac {n(k-2)}{2}$. Despite significant interest, the conjecture remains unsolved. Recently, Caro, Patkós, and Tuza considered this problem for host graphs that are connected. Settling a problem posed by them, for a $k$-vertex tree $T$, we construct $n$-vertex connected graphs that are $T$-free with at least $(1/4-o_k(1))nk$ edges, showing that the additional connectivity condition can reduce the maximum size by at most a factor of 2. Furthermore, we show that this is optimal: there is a family of $k$-vertex brooms $T$ such that the maximum size of an $n$-vertex connected $T$-free graph is at most $(1/4+o_k(1))nk$.

math.CO↗

arXiv2024

A Survey on 3D Skeleton-Based Action Recognition Using Learning Method

3D skeleton-based action recognition (3D SAR) has gained significant attention within the computer vision community, owing to the inherent advantages offered by skeleton data. As a result, a plethora of impressive works, including those based on conventional handcrafted features and learned feature extraction methods, have been conducted over the years. However, prior surveys on action recognition have primarily focused on video or RGB data-dominated approaches, with limited coverage of reviews related to skeleton data. Furthermore, despite the extensive application of deep learning methods in this field, there has been a notable absence of research that provides an introductory or comprehensive review from the perspective of deep learning architectures. To address these limitations, this survey first underscores the importance of action recognition and emphasizes the significance of 3D skeleton data as a valuable modality. Subsequently, we provide a comprehensive introduction to mainstream action recognition techniques based on four fundamental deep architectures, i.e., Recurrent Neural Networks (RNNs), Convolutional Neural Networks (CNNs), Graph Convolutional Network (GCN), and Transformers. All methods with the corresponding architectures are then presented in a data-driven manner with detailed discussion. Finally, we offer insights into the current largest 3D skeleton dataset, NTU-RGB+D, and its new edition, NTU-RGB+D 120, along with an overview of several top-performing algorithms on these datasets. To the best of our knowledge, this research represents the first comprehensive discussion of deep learning-based action recognition using 3D skeleton data.

cs.CV↗

arXiv2024

Dual-Modal Attention-Enhanced Text-Video Retrieval with Triplet Partial Margin Contrastive Learning

In recent years, the explosion of web videos makes text-video retrieval increasingly essential and popular for video filtering, recommendation, and search. Text-video retrieval aims to rank relevant text/video higher than irrelevant ones. The core of this task is to precisely measure the cross-modal similarity between texts and videos. Recently, contrastive learning methods have shown promising results for text-video retrieval, most of which focus on the construction of positive and negative pairs to learn text and video representations. Nevertheless, they do not pay enough attention to hard negative pairs and lack the ability to model different levels of semantic similarity. To address these two issues, this paper improves contrastive learning using two novel techniques. First, to exploit hard examples for robust discriminative power, we propose a novel Dual-Modal Attention-Enhanced Module (DMAE) to mine hard negative pairs from textual and visual clues. By further introducing a Negative-aware InfoNCE (NegNCE) loss, we are able to adaptively identify all these hard negatives and explicitly highlight their impacts in the training loss. Second, our work argues that triplet samples can better model fine-grained semantic similarity compared to pairwise samples. We thereby present a new Triplet Partial Margin Contrastive Learning (TPM-CL) module to construct partial order triplet samples by automatically generating fine-grained hard negatives for matched text-video pairs. The proposed TPM-CL designs an adaptive token masking strategy with cross-modal interaction to model subtle semantic differences. Extensive experiments demonstrate that the proposed approach outperforms existing methods on four widely-used text-video retrieval datasets, including MSR-VTT, MSVD, DiDeMo and ActivityNet.

cs.CV↗

arXiv2024

Dense Nuclear Matter Equation of State from Heavy-Ion Collisions

The nuclear equation of state (EOS) is at the center of numerous theoretical and experimental efforts in nuclear physics. With advances in microscopic theories for nuclear interactions, the availability of experiments probing nuclear matter under conditions not reached before, endeavors to develop sophisticated and reliable transport simulations to interpret these experiments, and the advent of multi-messenger astronomy, the next decade will bring new opportunities for determining the nuclear matter EOS, elucidating its dependence on density, temperature, and isospin asymmetry. Among controlled terrestrial experiments, collisions of heavy nuclei at intermediate beam energies (from a few tens of MeV/nucleon to about 25 GeV/nucleon in the fixed-target frame) probe the widest ranges of baryon density and temperature, enabling studies of nuclear matter from a few tenths to about 5 times the nuclear saturation density and for temperatures from a few to well above a hundred MeV, respectively. Collisions of neutron-rich isotopes further bring the opportunity to probe effects due to the isospin asymmetry. However, capitalizing on the enormous scientific effort aimed at uncovering the dense nuclear matter EOS, both at RHIC and at FRIB as well as at other international facilities, depends on the continued development of state-of-the-art hadronic transport simulations. This white paper highlights the essential role that heavy-ion collision experiments and hadronic transport simulations play in understanding strong interactions in dense nuclear matter, with an emphasis on how these efforts can be used together with microscopic approaches and neutron star studies to uncover the nuclear EOS.

nucl-th↗

arXiv2024

Volume Transfer: A New Design Concept for Fabric-Based Pneumatic Exosuits

The fabric-based pneumatic exosuit is now a hot research topic because it is lighter and softer than traditional exoskeletons. Existing research focused more on the mechanical properties of the exosuit (e.g., torque and speed), but less on its wearability (e.g., appearance and comfort). This work presents a new design concept for fabric-based pneumatic exosuits Volume Transfer, which means transferring the volume of pneumatic actuators beyond the garments profile to the inside. This allows for a concealed appearance and a larger stress area while maintaining adequate torques. In order to verify this concept, we develop a fabric-based pneumatic exosuit for knee extension assistance. Its profile is only 26mm and its stress area wraps around almost half of the leg. We use a mathematical model and simulation to determine the parameters of the exosuit, avoiding multiple iterations of the prototype. Experiment results show that the exosuit can generate a torque of 7.6Nm at a pressure of 90kPa and produce a significant reduction in the electromyography activity of the knee extensor muscles. We believe that Volume Transfer could be utilized prevalently in future fabric-based pneumatic exosuit designs to achieve a significant improvement in wearability.

cs.RO↗

arXiv2023

Audio Generation with Multiple Conditional Diffusion Model

Text-based audio generation models have limitations as they cannot encompass all the information in audio, leading to restricted controllability when relying solely on text. To address this issue, we propose a novel model that enhances the controllability of existing pre-trained text-to-audio models by incorporating additional conditions including content (timestamp) and style (pitch contour and energy contour) as supplements to the text. This approach achieves fine-grained control over the temporal order, pitch, and energy of generated audio. To preserve the diversity of generation, we employ a trainable control condition encoder that is enhanced by a large language model and a trainable Fusion-Net to encode and fuse the additional conditions while keeping the weights of the pre-trained text-to-audio model frozen. Due to the lack of suitable datasets and evaluation metrics, we consolidate existing datasets into a new dataset comprising the audio and corresponding conditions and use a series of evaluation metrics to evaluate the controllability performance. Experimental results demonstrate that our model successfully achieves fine-grained control to accomplish controllable audio generation. Audio samples and our dataset are publicly available at https://conditionaudiogen.github.io/conditionaudiogen/

cs.SD↗

arXiv2023

Algebraic ER=EPR and Complexity Transfer

We propose an algebraic definition of ER=EPR in the $G_N \to 0$ limit, which associates bulk spacetime connectivity/disconnectivity to the operator algebraic structure of a quantum gravity system. The new formulation not only includes information on the amount of entanglement, but also more importantly the structure of entanglement. We give an independent definition of a quantum wormhole as part of the proposal. This algebraic version of ER=EPR sheds light on a recent puzzle regarding spacetime disconnectivity in holographic systems with ${\cal O}(1/G_{N})$ entanglement. We discuss the emergence of quantum connectivity in the context of black hole evaporation and further argue that at the Page time, the black hole-radiation system undergoes a transition involving the transfer of an emergent type III$_{1}$ subalgebra of high complexity operators from the black hole to radiation. We argue this is a general phenomenon that occurs whenever there is an exchange of dominance between two competing quantum extremal surfaces.

hep-th↗

arXiv2023

New lower bounds on kissing numbers and spherical codes in high dimensions

Let the kissing number $K(d)$ be the maximum number of non-overlapping unit balls in $\mathbb R^d$ that can touch a given unit ball. Determining or estimating the number $K(d)$ has a long history, with the value of $K(3)$ being the subject of a famous discussion between Gregory and Newton in 1694. We prove that, as the dimension $d$ goes to infinity, $$ K(d)\ge (1+o(1)){\frac{\sqrt{3π}}{4\sqrt2}}\,\log\frac{3}{2}\cdot d^{3/2}\cdot \Big(\frac{2}{\sqrt{3}}\Big)^{d}, $$ thus improving the previously best known bound of Jenssen, Joos and Perkins by a factor of $\log(3/2)/\log(9/8)+o(1)=3.442...$. Our proof is based on the novel approach from Jenssen, Joos and Perkins that uses the hard core sphere model of an appropriate fugacity. Similar constant-factor improvements in lower bounds are also obtained for general spherical codes, as well as for the expected density of random sphere packings in the Euclidean space $\mathbb R^d$.

math.CO↗