Search

Most knowledge distillation (KD) methodologies predominantly focus on teacher-student pairs with similar architectures, such as both being convolutional neural networks (CNNs). However, the potential and flexibility of KD can be greatly improved by expanding it to novel Cross-Architecture KD (CAKD), where the knowledge of homogeneous and heterogeneous teachers can be transferred flexibly to a given student. The primary challenge in CAKD lies in the substantial feature gaps between heterogeneous models, originating from the distinction of their inherent inductive biases and module functions. To this end, we introduce an assistant model as a bridge to facilitate smooth feature knowledge transfer between heterogeneous teachers and students. More importantly, within our proposed design principle, the assistant model combines the advantages of cross-architecture inductive biases and module functions by merging convolution and attention modules derived from both student and teacher module functions. Furthermore, we observe that heterogeneous features exhibit diverse spatial distributions in CAKD, hindering the effectiveness of conventional pixel-wise mean squared error (MSE) loss. Therefore, we leverage a spatial-agnostic InfoNCE loss to align features after spatial smoothing, thereby improving the feature alignments in CAKD. Our proposed method is evaluated across some homogeneous model pairs and arbitrary heterogeneous combinations of CNNs, ViTs, and MLPs, achieving state-of-the-art performance for distilled models with a maximum gain of 11.47% on CIFAR-100 and 3.67% on ImageNet-1K. Our code and models will be released., Comment: 18 pages, 6 figures, and 12 tables

To alleviate hardware scarcity in training large deep neural networks (DNNs), particularly large language models (LLMs), we present FusionLLM, a decentralized training system designed and implemented for training DNNs using geo-distributed GPUs across different computing clusters or individual devices. Decentralized training faces significant challenges regarding system design and efficiency, including: 1) the need for remote automatic differentiation (RAD), 2) support for flexible model definitions and heterogeneous software, 3) heterogeneous hardware leading to low resource utilization or the straggler problem, and 4) slow network communication. To address these challenges, in the system design, we represent the model as a directed acyclic graph of operators (OP-DAG). Each node in the DAG represents the operator in the DNNs, while the edge represents the data dependency between operators. Based on this design, 1) users are allowed to customize any DNN without caring low-level operator implementation; 2) we enable the task scheduling with the more fine-grained sub-tasks, offering more optimization space; 3) a DAG runtime executor can implement RAD withour requiring the consistent low-level ML framework versions. To enhance system efficiency, we implement a workload estimator and design an OP-Fence scheduler to cluster devices with similar bandwidths together and partition the DAG to increase throughput. Additionally, we propose an AdaTopK compressor to adaptively compress intermediate activations and gradients at the slowest communication links. To evaluate the convergence and efficiency of our system and algorithms, we train ResNet-101 and GPT-2 on three real-world testbeds using 48 GPUs connected with 8 Mbps~10 Gbps networks. Experimental results demonstrate that our system and method can achieve 1.45 - 9.39x speedup compared to baseline methods while ensuring convergence.

Existing prompt learning methods in Vision-Language Models (VLM) have effectively enhanced the transfer capability of VLM to downstream tasks, but they suffer from a significant decline in generalization due to severe overfitting. To address this issue, we propose a framework named LOBG for vision-language models. Specifically, we use CLIP to filter out fine-grained foreground information that might cause overfitting, thereby guiding prompts with basic visual concepts. To further mitigate overfitting, we devel oped a structural topology preservation (STP) loss at the feature level, which endows the feature space with overall plasticity, allowing effective reshaping of the feature space during optimization. Additionally, we employed hierarchical logit distilation (HLD) at the output level to constrain outputs, complementing STP at the output end. Extensive experimental results demonstrate that our method significantly improves generalization capability and alleviates overfitting compared to state-of-the-art approaches.

In spite of the outstanding performance, Neural Architecture Search (NAS) is criticized for massive computation. Recently, Zero-shot NAS has emerged as a promising approach by exploiting Zero-cost (ZC) proxies, which markedly reduce computational demands. Despite this, existing ZC proxies heavily rely on expert knowledge and incur significant trial-and-error costs. Particularly in NLP tasks, most existing ZC proxies fail to surpass the performance of the naive baseline. To address these challenges, we introduce a novel framework, \textbf{LPZero}, which is the first to automatically design ZC proxies for various tasks, achieving higher ranking consistency than human-designed proxies. Specifically, we model the ZC proxy as a symbolic equation and incorporate a unified proxy search space that encompasses existing ZC proxies, which are composed of a predefined set of mathematical symbols. To heuristically search for the best ZC proxy, LPZero incorporates genetic programming to find the optimal symbolic composition. We propose a \textit{Rule-based Pruning Strategy (RPS),} which preemptively eliminates unpromising proxies, thereby mitigating the risk of proxy degradation. Extensive experiments on FlexiBERT, GPT-2, and LLaMA-7B demonstrate LPZero's superior ranking ability and performance on downstream tasks compared to current approaches., Comment: 8 pages, 7 figures, 10 appendix

Lifelong person re-identification (LReID) aims to continuously learn from non-stationary data to match individuals in different environments. Each task is affected by variations in illumination and person-related information (such as pose and clothing), leading to task-wise domain gaps. Current LReID methods focus on task-specific knowledge and ignore intrinsic task-shared representations within domain gaps, limiting model performance. Bridging task-wise domain gaps is crucial for improving anti-forgetting and generalization capabilities, especially when accessing limited old classes during training. To address these issues, we propose a novel attribute-text guided forgetting compensation (ATFC) model, which explores text-driven global representations of identity-related information and attribute-related local representations of identity-free information for LReID. Due to the lack of paired text-image data, we design an attribute-text generator (ATG) to dynamically generate a text descriptor for each instance. We then introduce a text-guided aggregation network (TGA) to explore robust text-driven global representations for each identity and knowledge transfer. Furthermore, we propose an attribute compensation network (ACN) to investigate attribute-related local representations, which distinguish similar identities and bridge domain gaps. Finally, we develop an attribute anti-forgetting (AF) loss and knowledge transfer (KT) loss to minimize domain gaps and achieve knowledge transfer, improving model performance. Extensive experiments demonstrate that our ATFC method achieves superior performance, outperforming existing LReID methods by over 9.0$\%$/7.4$\%$ in average mAP/R-1 on the seen dataset., Comment: 9 pages, 4 figures

The recent experiment shows the superconducting transition temperature in the Ruddlesden-Popper bilayer La$_3$Ni$_2$O$_{7}$ decreases monotonically with increasing pressure above 14 GPa. In order to unravel the underlying mechanism for this unusual dependence, we performed theoretical investigations by combining the density functional theory (DFT) and the unbiased functional renormalization group (FRG). Our DFT calculations show that the Fermi pockets are essentially unchanged with increasing pressure (above 14 GPa), but the bandwidth is enlarged, and particularly the interlayer hopping integral between the nickel $3d_{3z^2-r^2}$ orbitals is enhanced. From the DFT band structure, we construct the bilayer tight-binding model in terms of the nickel $3d_{3z^2-r^2}$ and $3d_{x^2-y^2}$ orbitals. On this basis, we investigate the superconductivity induced by correlation effects by FRG calculations. We find consistently $s_\pm$-wave pairing triggered by spin fluctuations, but the latter are weakened by pressure and lead to a decreasing transition temperature versus pressure, in qualitatively agreement with the experiment. We emphasize that the itinerancy of the $d$-orbitals is important and captured naturally in our FRG calculations, and we argue that the unusual pressure dependence would be unnatural, if not impossible, in the otherwise local-moment picture of the nickel $d$-orbitals. This sheds lights on the pertinent microscopic description of, and more importantly the mechanism of superconductivity in La$_3$Ni$_2$O$_{7}$., Comment: 8 pages, 4 figures

The definition of many-to-one mapping, or $m$-to-$1$ mapping for short, between two finite sets is introduced in this paper, which unifies and generalizes the definitions of $2$-to-$1$ mappings and $n$-to-$1$ mappings. A generalized local criterion is given, which is an abstract criterion for a mapping to be $m$-to-$1$. By employing the generalized local criterion, three constructions of $m$-to-$1$ mapping are proposed, which unify and generalize all the previous constructions of $2$-to-$1$ mappings and $n$-to-$1$ mappings. Then the $m$-to-$1$ property of polynomials $f(x) = x^r h(x^s)$ on $\mathbb{F}_{q}^{*}$ is studied by using these three constructions. A series of explicit conditions for~$f$ to be an $m$-to-$1$ mapping on $\mathbb{F}_{q}^{*}$ are found through the detailed discussion of the parameters $m$, $s$, $q$ and the polynomial $h$. These results extend many conclusions in the literature.

In challenging low light and adverse weather conditions,thermal vision algorithms,especially object detection,have exhibited remarkable potential,contrasting with the frequent struggles encountered by visible vision algorithms. Nevertheless,the efficacy of thermal vision algorithms driven by deep learning models remains constrained by the paucity of available training data samples. To this end,this paper introduces a novel approach termed the edge guided conditional diffusion model. This framework aims to produce meticulously aligned pseudo thermal images at the pixel level,leveraging edge information extracted from visible images. By utilizing edges as contextual cues from the visible domain,the diffusion model achieves meticulous control over the delineation of objects within the generated images. To alleviate the impacts of those visible-specific edge information that should not appear in the thermal domain,a two-stage modality adversarial training strategy is proposed to filter them out from the generated images by differentiating the visible and thermal modality. Extensive experiments on LLVIP demonstrate ECDM s superiority over existing state-of-the-art approaches in terms of image generation quality., Comment: accepted by ACM MM 2024/ACM MM24

Pair density wave (PDW) is an intriguing quantum matter proposed in the frontier of condensed matter physics. However, the existence of PDW in microscopic models has been rare. In this work, we obtain, by Ginzburg-Landau arguments and self-consistent mean field theory, novel $2a_0\times2a_0$ PDW on the kagome lattice arising from attractive on-bond pairing interactions and the distinct Bloch wave functions near the p-type van Hove singularity. The PDW state carrying three independent wave-vectors, the so-called 3Q PDW, is nodeless and falls into two topological classes characterized by the Chern number $C = 0$ or $C = \pm2$. The chiral ($C=\pm2$) PDW state presents a rare case of interaction driven topological quantum state without the requirement of spin-orbit coupling. Finally, we analyze the stabilities and properties of these PDWs intertwining with charge orders, and discuss the relevance of our minimal model to recent experimental observations in kagome superconductors. Our theory not only elucidates the driving force of the chiral PDW, but also predicts strongly anisotropic superconducting gap structure in the momentum space and quantized transverse thermal conductivity that can be tested in future experiments., Comment: 7 pages, 4 figures

In this paper, we present the first structural binarization method for LLM compression to less than 1-bit precision. Although LLMs have achieved remarkable performance, their memory-bound nature during the inference stage hinders the adoption of resource-constrained devices. Reducing weights to 1-bit precision through binarization substantially enhances computational efficiency. We observe that some weights in binarized LLMs can be randomly flipped without significant performance degradation, suggesting the potential for further compression. To exploit this, our STBLLM employs an N:M sparsity technique to achieve structural binarization of the weights. Specifically, we introduce a novel Standardized Importance (SI) metric, which considers weight magnitude and input feature norm to more accurately assess weight significance. Then, we propose a layer-wise approach, allowing different layers of the LLM to be sparsified with varying N:M ratios, thereby balancing compression and accuracy. Furthermore, we implement a fine-grained grouping strategy for less important weights, applying distinct quantization schemes to sparse, intermediate, and dense regions. Finally, we design a specialized CUDA kernel to support structural binarization. We conduct extensive experiments on LLaMA-1/2/3, OPT family, and Mistral to evaluate the effectiveness of STBLLM. The results demonstrate that our approach performs better than other compressed binarization LLM methods while significantly reducing memory requirements.

Experimental results show that, when the order $n$ is odd, there are de Bruijn sequences such that the corresponding complement sequence and the reverse sequence are the same. In this paper, we propose one efficient method to generate such de Bruijn sequences. This solves an open problem asked by Fredricksen forty years ago for showing the existence of such de Bruijn sequences when the odd order $n >1$. Moreover, we refine a characterization of de Bruijn sequences with the same complement and reverse sequences and study the number of these de Bruijn sequences, as well as the distribution of de Bruijn sequences of the maximum linear complexity.

3D multimodal question answering (MQA) plays a crucial role in scene understanding by enabling intelligent agents to comprehend their surroundings in 3D environments. While existing research has primarily focused on indoor household tasks and outdoor roadside autonomous driving tasks, there has been limited exploration of city-level scene understanding tasks. Furthermore, existing research faces challenges in understanding city scenes, due to the absence of spatial semantic information and human-environment interaction information at the city level.To address these challenges, we investigate 3D MQA from both dataset and method perspectives. From the dataset perspective, we introduce a novel 3D MQA dataset named City-3DQA for city-level scene understanding, which is the first dataset to incorporate scene semantic and human-environment interactive tasks within the city. From the method perspective, we propose a Scene graph enhanced City-level Understanding method (Sg-CityU), which utilizes the scene graph to introduce the spatial semantic. A new benchmark is reported and our proposed Sg-CityU achieves accuracy of 63.94 % and 63.76 % in different settings of City-3DQA. Compared to indoor 3D MQA methods and zero-shot using advanced large language models (LLMs), Sg-CityU demonstrates state-of-the-art (SOTA) performance in robustness and generalization.

Large Language Models (LLMs) have demonstrated impressive performance across various domains. However, the enormous number of model parameters makes fine-tuning challenging, significantly limiting their application and deployment. Existing solutions combine parameter quantization with Low-Rank Adaptation (LoRA), greatly reducing memory usage but resulting in noticeable performance degradation. In this paper, we identify an imbalance in fine-tuning quantized pre-trained models: overly complex adapter inputs and outputs versus low effective trainability of the adaptation. We propose Quantized LLMs with Balanced-rank Adaptation (Q-BaRA), which simplifies the adapter inputs and outputs while increasing the adapter's rank to achieve a more suitable balance for fine-tuning quantized LLMs. Additionally, for scenarios where fine-tuned LLMs need to be deployed as low-precision inference models, we introduce Quantization-Aware Fine-tuning with Higher Rank Adaptation (QA-HiRA), which simplifies the adapter inputs and outputs to align with the pre-trained model's block-wise quantization while employing a single matrix to achieve a higher rank. Both Q-BaRA and QA-HiRA are easily implemented and offer the following optimizations: (i) Q-BaRA consistently achieves the highest accuracy compared to baselines and other variants, requiring the same number of trainable parameters and computational effort; (ii) QA-HiRA naturally merges adapter parameters into the block-wise quantized model after fine-tuning, achieving the highest accuracy compared to other methods. We apply our Q-BaRA and QA-HiRA to the LLaMA and LLaMA2 model families and validate their effectiveness across different fine-tuning datasets and downstream scenarios. Code will be made available at \href{https://github.com/xiaocaigou/qbaraqahira}{https://github.com/xiaocaigou/qbaraqahira}

Increased reliance on graphics processing units (GPUs) for high-intensity computing tasks raises challenges regarding energy consumption. To address this issue, dynamic voltage and frequency scaling (DVFS) has emerged as a promising technique for conserving energy while maintaining the quality of service (QoS) of GPU applications. However, existing solutions using DVFS are hindered by inefficiency or inaccuracy as they depend either on dynamic or static information respectively, which prevents them from being adopted to practical power management schemes. To this end, we propose a novel energy efficiency optimizer, called DSO, to explore a light weight solution that leverages both dynamic and static information to model and optimize the GPU energy efficiency. DSO firstly proposes a novel theoretical energy efficiency model which reflects the DVFS roofline phenomenon and considers the tradeoff between performance and energy. Then it applies machine learning techniques to predict the parameters of the above model with both GPU kernel runtime metrics and static code features. Experiments on modern DVFS-enabled GPUs indicate that DSO can enhance energy efficiency by 19% whilst maintaining performance within a 5% loss margin.

Deep learning (DL) has demonstrated significant success across diverse fields, leading to the construction of dedicated GPU accelerators within GPU clusters for high-quality training services. Efficient scheduler designs for such clusters are vital to reduce operational costs and enhance resource utilization. While recent schedulers have shown impressive performance in optimizing DL job performance and cluster utilization through periodic reallocation or selection of GPU resources, they also encounter challenges such as preemption and migration overhead, along with potential DL accuracy degradation. Nonetheless, few explore the potential benefits of GPU sharing to improve resource utilization and reduce job queuing times. Motivated by these insights, we present a job scheduling model allowing multiple jobs to share the same set of GPUs without altering job training settings. We introduce SJF-BSBF (shortest job first with best sharing benefit first), a straightforward yet effective heuristic scheduling algorithm. SJF-BSBF intelligently selects job pairs for GPU resource sharing and runtime settings (sub-batch size and scheduling time point) to optimize overall performance while ensuring DL convergence accuracy through gradient accumulation. In experiments with both physical DL workloads and trace-driven simulations, even as a preemption-free policy, SJF-BSBF reduces the average job completion time by 27-33\% relative to the state-of-the-art preemptive DL schedulers. Moreover, SJF-BSBF can wisely determine the optimal resource sharing settings, such as the sharing time point and sub-batch size for gradient accumulation, outperforming the aggressive GPU sharing approach (baseline SJF-FFS policy) by up to 17\% in large-scale traces.

The problem of Rehearsal-Free Continual Learning (RFCL) aims to continually learn new knowledge while preventing forgetting of the old knowledge, without storing any old samples and prototypes. The latest methods leverage large-scale pre-trained models as the backbone and use key-query matching to generate trainable prompts to learn new knowledge. However, the domain gap between the pre-training dataset and the downstream datasets can easily lead to inaccuracies in key-query matching prompt selection when directly generating queries using the pre-trained model, which hampers learning new knowledge. Thus, in this paper, we propose a beyond prompt learning approach to the RFCL task, called Continual Adapter (C-ADA). It mainly comprises a parameter-extensible continual adapter layer (CAL) and a scaling and shifting (S&S) module in parallel with the pre-trained model. C-ADA flexibly extends specific weights in CAL to learn new knowledge for each task and freezes old weights to preserve prior knowledge, thereby avoiding matching errors and operational inefficiencies introduced by key-query matching. To reduce the gap, C-ADA employs an S&S module to transfer the feature space from pre-trained datasets to downstream datasets. Moreover, we propose an orthogonal loss to mitigate the interaction between old and new knowledge. Our approach achieves significantly improved performance and training speed, outperforming the current state-of-the-art (SOTA) method. Additionally, we conduct experiments on domain-incremental learning, surpassing the SOTA, and demonstrating the generality of our approach in different settings., Comment: ECCV2024

Vortex is a topological defect in the superconducting condensate when a magnetic field is applied to a type-II superconductor, as elucidated by the Ginzburg-Landau theory. Due to the confinement of the quasiparticles by a vortex, it exhibits a circular shaped pattern of bound states with discrete energy levels, as predicted by the Caroli-de Gennes-Matricon theory in 1964. Here, however, we report a completely new type of vortex pattern which is necklace-like in an iron-based superconductor KCa2Fe4As4F2. Our theoretical analysis shows that this necklace-like vortex pattern arises from selective off-shell interference between vortex bound states of opposite angular momenta in the presence of rotational symmetry breaking due to disorders. This fascinating effect can be observed in a system with a small Fermi energy and wave vector, conditions fortuitously met in our samples. Our results not only disclose a novel vortex structure but also provide insights into comprehending the physics of the superconducting condensate., Comment: 29 pages total; 16 pages of main text with 5 figures, 13 pages of supplementary materials with 10 figures

Travel time estimation is a crucial application in navigation apps and web mapping services. Current deterministic and probabilistic methods primarily focus on modeling individual trips, assuming independence among trips. However, in real-world scenarios, we often observe strong inter-trip correlations due to factors such as weather conditions, traffic management, and road works. In this paper, we propose to model trip-level link travel time using a Gaussian hierarchical model, which can characterize both inter-trip and intra-trip correlations. The joint distribution of travel time of multiple trips becomes a multivariate Gaussian parameterized by learnable link representations. To effectively use the sparse GPS trajectories, we also propose a data augmentation method based on trip sub-sampling, which allows for fine-grained gradient backpropagation in learning link representations. During inference, we estimate the probability distribution of the travel time of a queried trip conditional on the completed trips that are spatiotemporally adjacent. We refer to the overall framework as ProbTTE. We evaluate ProbTTE on two real-world GPS trajectory datasets, and the results demonstrate its superior performance compared to state-of-the-art deterministic and probabilistic baselines. Additionally, we find that the learned link representations align well with the physical geometry of the network, making them suitable as input for other applications.

The existing works on object-level language grounding with 3D objects mostly focus on improving performance by utilizing the off-the-shelf pre-trained models to capture features, such as viewpoint selection or geometric priors. However, they have failed to consider exploring the cross-modal representation of language-vision alignment in the cross-domain field. To answer this problem, we propose a novel method called Domain Adaptation for Language Grounding (DA4LG) with 3D objects. Specifically, the proposed DA4LG consists of a visual adapter module with multi-task learning to realize vision-language alignment by comprehensive multimodal feature representation. Experimental results demonstrate that DA4LG competitively performs across visual and non-visual language descriptions, independent of the completeness of observation. DA4LG achieves state-of-the-art performance in the single-view setting and multi-view setting with the accuracy of 83.8% and 86.8% respectively in the language grounding benchmark SNARE. The simulation experiments show the well-practical and generalized performance of DA4LG compared to the existing methods. Our project is available at https://sites.google.com/view/da4lg.

In the task of multi-target multi-camera (MTMC) tracking of pedestrians, the data association problem is a key issue and main challenge, especially with complications arising from camera movements, lighting variations, and obstructions. However, most MTMC models adopt two-step approaches, thus heavily depending on the results of the first-step tracking in practical applications. Moreover, the same targets crossing different cameras may exhibit significant appearance variations, which further increases the difficulty of cross-camera matching. To address the aforementioned issues, we propose a global online MTMC tracking model that addresses the dependency on the first tracking stage in two-step methods and enhances cross-camera matching. Specifically, we propose a transformer-based global MTMC association module to explore target associations across different cameras and frames, generating global trajectories directly. Additionally, to integrate the appearance and spatio-temporal features of targets, we propose a feature extraction and fusion module for MTMC tracking. This module enhances feature representation and establishes correlations between the features of targets across multiple cameras. To accommodate high scene diversity and complex lighting condition variations, we have established the VisionTrack dataset, which enables the development of models that are more generalized and robust to various environments. Our model demonstrates significant improvements over comparison methods on the VisionTrack dataset and others.

Driving systems often rely on high-definition (HD) maps for precise environmental information, which is crucial for planning and navigation. While current HD map constructors perform well under ideal conditions, their resilience to real-world challenges, \eg, adverse weather and sensor failures, is not well understood, raising safety concerns. This work introduces MapBench, the first comprehensive benchmark designed to evaluate the robustness of HD map construction methods against various sensor corruptions. Our benchmark encompasses a total of 29 types of corruptions that occur from cameras and LiDAR sensors. Extensive evaluations across 31 HD map constructors reveal significant performance degradation of existing methods under adverse weather conditions and sensor failures, underscoring critical safety concerns. We identify effective strategies for enhancing robustness, including innovative approaches that leverage multi-modal fusion, advanced data augmentation, and architectural techniques. These insights provide a pathway for developing more reliable HD map construction methods, which are essential for the advancement of autonomous driving technology. The benchmark toolkit and affiliated code and model checkpoints have been made publicly accessible., Comment: NeurIPS 2024; 40 pages, 17 figures, 23 tables; Code at https://mapbench.github.io/

Despite the remarkable capabilities, Large Language Models (LLMs) face deployment challenges due to their extensive size. Pruning methods drop a subset of weights to accelerate, but many of them require retraining, which is prohibitively expensive and computationally demanding. Recently, post-training pruning approaches introduced novel metrics, enabling the pruning of LLMs without retraining. However, these metrics require the involvement of human experts and tedious trial and error. To efficiently identify superior pruning metrics, we develop an automatic framework for searching symbolic pruning metrics using genetic programming. In particular, we devise an elaborate search space encompassing the existing pruning metrics to discover the potential symbolic pruning metric. We propose an opposing operation simplification strategy to increase the diversity of the population. In this way, Pruner-Zero allows auto-generation of symbolic pruning metrics. Based on the searched results, we explore the correlation between pruning metrics and performance after pruning and summarize some principles. Extensive experiments on LLaMA and LLaMA-2 on language modeling and zero-shot tasks demonstrate that our Pruner-Zero obtains superior performance than SOTA post-training pruning methods. Code at: \url{https://github.com/pprp/Pruner-Zero}., Comment: Accepted by ICML2024, 29 pages, 4 figures

Among the most intriguing features of non-Hermitian (NH) systems is the ability of complex energies to form braids under parametric variation. Several braiding behaviors, including link and knot formation, have been observed in experiments on synthetic NH systems, such as looped optical fibers. The exact conditions for these phenomena remain unsettled, but existing demonstrations have involved long-range nonreciprocal hoppings, which are hard to implement on many experimental platforms. Here, we present a route to realizing complex energy braids using 1D NH Aubry-Andr\'e-Harper lattices. Under purely local gain and loss modulation, the eigenstates exhibit a variety of braiding behaviors, including unknots, Hopf links, trefoil knots, Solomon links and catenanes. We show how these are created by the interplay between non-Hermiticity and the lattice's bulk states and topological edge states. The transitions between different braids are marked by changes in the global Berry phase of the NH lattice.

The scarcity of free-hand sketch presents a challenging problem. Despite the emergence of some large-scale sketch datasets, these datasets primarily consist of sketches at the single-object level. There continues to be a lack of large-scale paired datasets for scene sketches. In this paper, we propose a self-supervised method for scene sketch generation that does not rely on any existing scene sketch, enabling the transformation of single-object sketches into scene sketches. To accomplish this, we introduce a method for vector sketch captioning and sketch semantic expansion. Additionally, we design a sketch generation network that incorporates a fusion of multi-modal perceptual constraints, suitable for application in zero-shot image-to-sketch downstream task, demonstrating state-of-the-art performance through experimental validation. Finally, leveraging our proposed sketch-to-sketch generation method, we contribute a large-scale dataset centered around scene sketches, comprising highly semantically consistent "text-sketch-image" triplets. Our research confirms that this dataset can significantly enhance the capabilities of existing models in sketch-based image retrieval and sketch-controlled image synthesis tasks. We will make our dataset and code publicly available.

Generating realistic animated videos from static images is an important area of research in computer vision. Methods based on physical simulation and motion prediction have achieved notable advances, but they are often limited to specific object textures and motion trajectories, failing to exhibit highly complex environments and physical dynamics. In this paper, we introduce an open-domain controllable image animation method using motion priors with video diffusion models. Our method achieves precise control over the direction and speed of motion in the movable region by extracting the motion field information from videos and learning moving trajectories and strengths. Current pretrained video generation models are typically limited to producing very short videos, typically less than 30 frames. In contrast, we propose an efficient long-duration video generation method based on noise reschedule specifically tailored for image animation tasks, facilitating the creation of videos over 100 frames in length while maintaining consistency in content scenery and motion coordination. Specifically, we decompose the denoise process into two distinct phases: the shaping of scene contours and the refining of motion details. Then we reschedule the noise to control the generated frame sequences maintaining long-distance noise correlation. We conducted extensive experiments with 10 baselines, encompassing both commercial tools and academic methodologies, which demonstrate the superiority of our method. Our project page: https://wangqiang9.github.io/Controllable.github.io/, Comment: https://wangqiang9.github.io/Controllable.github.io/

Isotope effect with a large coefficient $\alpha=-\partial \ln T_c/\partial \ln M$ is usually taken as an evidence of phonon mediated superconductors in the Bardeen-Cooper-Schrieffer (BCS) theory. However, in cuprates which are now widely believed to be strong correlation induced d-wave superconductors, $\alpha$ is experimentally observed to be quite small at optimal doping, but keeps growing up with decreasing $T_c$ upon doping, even after exceeding the BCS value $1/2$. Such an anomalous isotope effect seems to challenge the non-phonon picture and still leave room for the phonon-dominated mechanism. In this work, we show that the anomalous dependence of $\alpha$ on $T_c$ can actually be obtained in spin fluctuation induced d-wave superconductors, by studying the Hubbard model on square lattices with functional renormalization group. We have considered two types of electron-phonon couplings (EPCs). The first type couples to electron densities, including the Holstein, breathing and buckling phonons, called Holstein-like. For all these EPCs, $\alpha$ is negative and drops down towards $-\infty$ with decreasing $T_c$ upon doping. On the opposite, for the other type of Peierls-like EPC coupling to electron hoppings on the nearest bonds, also called Su-Schrieffer-Heeger phonon, $\alpha$ is positive, grows up with decreasing $T_c$ and tends to diverge as $T_c\to0$, in qualitative agreement with the experiments. The difference between these two types of EPCs can be understood by their isotope effects on spin fluctuations. From this study, we conclude that the SSH phonon can explain the anomalous isotope effect in cuprates, although it is not the leading pairing mechanism.

The bound states around a vortex in anisotropic superconductors is a longstanding yet important issue. In this work, we develop a variational theory on the basis of the Andreev approximation to obtain the energy levels and wave functions of the low-energy quantized bound states in superconductors with anisotropic pairing on arbitrary Fermi surface. In the case of circular Fermi surface, the effective Schr\"odinger equation yielding the bound state energies gets back to the theory proposed by Volovik and Kopnin many years ago. Our generalization here enables us to prove the equidistant energy spectrum inside a vortex in a broader class of superconductors. More importantly, we are now able to obtain the wave functions of these bound states by projecting the quasiclassical wave function on the eigenmodes of the effective Schr\"odinger equation, going beyond the quasiclassical Eilenberger results, which, as we find, are sensitive to the scattering rate. For the case of isotropic Fermi surface, the spatial profile of the low-energy local density of states is dominated near the vortex center and elongates along the gap antinode directions, in addition to the ubiquitous Friedel oscillation arising from the quantum inteference neglected in the Eilenberger theory. Moreover, as a consequence of the pairing anisotropy, the quantized wave functions develop a peculiar distribution of winding number, which reduces stepwise towards the vortex center. Our work provides a flexible way to study the vortex bound states in the future.

Point scene understanding is a challenging task to process real-world scene point cloud, which aims at segmenting each object, estimating its pose, and reconstructing its mesh simultaneously. Recent state-of-the-art method first segments each object and then processes them independently with multiple stages for the different sub-tasks. This leads to a complex pipeline to optimize and makes it hard to leverage the relationship constraints between multiple objects. In this work, we propose a novel Disentangled Object-Centric TRansformer (DOCTR) that explores object-centric representation to facilitate learning with multiple objects for the multiple sub-tasks in a unified manner. Each object is represented as a query, and a Transformer decoder is adapted to iteratively optimize all the queries involving their relationship. In particular, we introduce a semantic-geometry disentangled query (SGDQ) design that enables the query features to attend separately to semantic information and geometric information relevant to the corresponding sub-tasks. A hybrid bipartite matching module is employed to well use the supervisions from all the sub-tasks during training. Qualitative and quantitative experimental results demonstrate that our method achieves state-of-the-art performance on the challenging ScanNet dataset. Code is available at https://github.com/SAITPublic/DOCTR.

Lifelong Person Re-Identification (LReID) aims to continuously learn from successive data streams, matching individuals across multiple cameras. The key challenge for LReID is how to effectively preserve old knowledge while incrementally learning new information, which is caused by task-level domain gaps and limited old task datasets. Existing methods based on CNN backbone are insufficient to explore the representation of each instance from different perspectives, limiting model performance on limited old task datasets and new task datasets. Unlike these methods, we propose a Diverse Representations Embedding (DRE) framework that first explores a pure transformer for LReID. The proposed DRE preserves old knowledge while adapting to new information based on instance-level and task-level layout. Concretely, an Adaptive Constraint Module (ACM) is proposed to implement integration and push away operations between multiple overlapping representations generated by transformer-based backbone, obtaining rich and discriminative representations for each instance to improve adaptive ability of LReID. Based on the processed diverse representations, we propose Knowledge Update (KU) and Knowledge Preservation (KP) strategies at the task-level layout by introducing the adjustment model and the learner model. KU strategy enhances the adaptive learning ability of learner models for new information under the adjustment model prior, and KP strategy preserves old knowledge operated by representation-level alignment and logit-level supervision in limited old task datasets while guaranteeing the adaptive learning information capacity of the LReID model. Compared to state-of-the-art methods, our method achieves significantly improved performance in holistic, large-scale, and occluded datasets., Comment: 11 pages,7 Tables,3 Figures

Depth-based 3D hand pose estimation is an important but challenging research task in human-machine interaction community. Recently, dense regression methods have attracted increasing attention in 3D hand pose estimation task, which provide a low computational burden and high accuracy regression way by densely regressing hand joint offset maps. However, large-scale regression offset values are often affected by noise and outliers, leading to a significant drop in accuracy. To tackle this, we re-formulate 3D hand pose estimation as a dense ordinal regression problem and propose a novel Dense Ordinal Regression 3D Pose Network (DOR3D-Net). Specifically, we first decompose offset value regression into sub-tasks of binary classifications with ordinal constraints. Then, each binary classifier can predict the probability of a binary spatial relationship relative to joint, which is easier to train and yield much lower level of noise. The estimated hand joint positions are inferred by aggregating the ordinal regression results at local positions with a weighted sum. Furthermore, both joint regression loss and ordinal regression loss are used to train our DOR3D-Net in an end-to-end manner. Extensive experiments on public datasets (ICVL, MSRA, NYU and HANDS2017) show that our design provides significant improvements over SOTA methods.