Search

Nonadiabatic holonomic quantum computation (NHQC) is one of the promising approaches to realizing fault-tolerant quantum computation. However, due to the imperfect control in the experimental environments, the holonomic gate still needs to be further improved. Here, we propose a dynamically optimized NHQC (OPNHQC) scheme based on dynamically corrected gate technique. The scheme is implemented by carefully designing a sequence of elementary pulses to fulfill cyclic evolution, while the dynamical phase is not accumulated. In this way, the constructed holonomic gate is immune to the error. It is found that our scheme can correct the $X$ error up to fourth order. In addition, combining with the DFS encoding our scheme can be immune to both the $X$ and $Z$ errors. Therefore, our proposed scheme offers a prospective way to the realization of scalable fault-tolerant holonomic quantum computation., Comment: 9 pages, 7 figures

This paper delves into recent hardware implementations of the Lempel-Ziv 4 (LZ4) algorithm, highlighting two key factors that limit the throughput of single-kernel compressors. Firstly, the actual parallelism exhibited in single-kernel designs falls short of the theoretical potential. Secondly, the clock frequency is constrained due to the presence of the feedback loops. To tackle these challenges, we propose a novel scheme that restricts each parallelization window to a single match, thus elevating the level of actual parallelism. Furthermore, by restricting the maximum match length, we eliminate the feedback loops within the architecture, enabling a significant boost in throughput. Finally, we present a high-speed hardware architecture. The implementation results demonstrate that the proposed architecture achieves a throughput of up to 16.10 Gb/s, exhibiting a 2.648x improvement over the start-of-the-art. The new design only results in an acceptable compression ratio reduction ranging from 4.93% to 11.68% with various numbers of hash table entries, compared to the LZ4 compression ratio achieved by official software implementations disclosed on GitHub.

the action of the function on its Julia set is still ergodic if some, but not all of the asymptotic values land on infinity, and the remaining ones land on a compact repeller. In this paper, we complete the characterization of ergodicity for Nevanlinna functions but proving that if all the asymptotic values land on infinity, then the Julia set is the whole sphere and the action of the map there is non-ergodic., Comment: 18 pages, 4 figures

Current Pose-Guided Person Image Synthesis (PGPIS) methods depend heavily on large amounts of labeled triplet data to train the generator in a supervised manner. However, they often falter when applied to in-the-wild samples, primarily due to the distribution gap between the training datasets and real-world test samples. While some researchers aim to enhance model generalizability through sophisticated training procedures, advanced architectures, or by creating more diverse datasets, we adopt the test-time fine-tuning paradigm to customize a pre-trained Text2Image (T2I) model. However, naively applying test-time tuning results in inconsistencies in facial identities and appearance attributes. To address this, we introduce a Visual Consistency Module (VCM), which enhances appearance consistency by combining the face, text, and image embedding. Our approach, named OnePoseTrans, requires only a single source image to generate high-quality pose transfer results, offering greater stability than state-of-the-art data-driven methods. For each test case, OnePoseTrans customizes a model in around 48 seconds with an NVIDIA V100 GPU.

Machine/deep learning models have been widely adopted for predicting the configuration performance of software systems. However, a crucial yet unaddressed challenge is how to cater for the sparsity inherited from the configuration landscape: the influence of configuration options (features) and the distribution of data samples are highly sparse. In this paper, we propose a model-agnostic and sparsity-robust framework for predicting configuration performance, dubbed DaL, based on the new paradigm of dividable learning that builds a model via "divide-and-learn". To handle sample sparsity, the samples from the configuration landscape are divided into distant divisions, for each of which we build a sparse local model, e.g., regularized Hierarchical Interaction Neural Network, to deal with the feature sparsity. A newly given configuration would then be assigned to the right model of division for the final prediction. Further, DaL adaptively determines the optimal number of divisions required for a system and sample size without any extra training or profiling. Experiment results from 12 real-world systems and five sets of training data reveal that, compared with the state-of-the-art approaches, DaL performs no worse than the best counterpart on 44 out of 60 cases with up to 1.61x improvement on accuracy; requires fewer samples to reach the same/better accuracy; and producing acceptable training overhead. In particular, the mechanism that adapted the parameter d can reach the optimal value for 76.43% of the individual runs. The result also confirms that the paradigm of dividable learning is more suitable than other similar paradigms such as ensemble learning for predicting configuration performance. Practically, DaL considerably improves different global models when using them as the underlying local models, which further strengthens its flexibility., Comment: Submitted to TSE as a regular journal paper. arXiv admin note: text overlap with arXiv:2306.06651

Disordered lattice spin systems are crucial in both theoretical and applied physics. However, understanding their properties poses significant challenges for Monte Carlo simulations. In this work, we investigate the two-dimensional random-bond Ising model using the recently proposed Tensor Network Monte Carlo (TNMC) method. This method generates biased samples from conditional probabilities computed via tensor network contractions and corrects the bias using the Metropolis scheme. Consequently, the proposals provided by tensor networks function as block updates for Monte Carlo simulations. Through extensive numerical experiments, we demonstrate that TNMC simulations can be performed on lattices as large as $1024\times 1024$ spins with moderate computational resources, a substantial increase from the previous maximum size of $64\times 64$ in MCMC. Notably, we observe an almost complete absence of critical slowing down, enabling the efficient collection of unbiased samples and averaging over a large number of random realizations of bond disorders. We successfully pinpoint the multi-critical point along the Nishimori line with significant precision and accurately determined the bulk and surface critical exponents. Our findings suggest that TNMC is a highly efficient algorithm for exploring disordered and frustrated systems in two dimensions., Comment: 14 pages, 11 figures

Dynamic scene reconstruction has garnered significant attention in recent years due to its capabilities in high-quality and real-time rendering. Among various methodologies, constructing a 4D spatial-temporal representation, such as 4D-GS, has gained popularity for its high-quality rendered images. However, these methods often produce suboptimal surfaces, as the discrete 3D Gaussian point clouds fail to align with the object's surface precisely. To address this problem, we propose DynaSurfGS to achieve both photorealistic rendering and high-fidelity surface reconstruction of dynamic scenarios. Specifically, the DynaSurfGS framework first incorporates Gaussian features from 4D neural voxels with the planar-based Gaussian Splatting to facilitate precise surface reconstruction. It leverages normal regularization to enforce the smoothness of the surface of dynamic objects. It also incorporates the as-rigid-as-possible (ARAP) constraint to maintain the approximate rigidity of local neighborhoods of 3D Gaussians between timesteps and ensure that adjacent 3D Gaussians remain closely aligned throughout. Extensive experiments demonstrate that DynaSurfGS surpasses state-of-the-art methods in both high-fidelity surface reconstruction and photorealistic rendering., Comment: homepage: https://open3dvlab.github.io/DynaSurfGS/, code: https://github.com/Open3DVLab/DynaSurfGS

Vorticity is locally created on a boundary at the rate measured by the boundary vorticity flux, which can be further decomposed as the sum of the orbital rotation and the (generalized) spin. For incompressible viscous flow interacting with a stationary wall, the full expressions of the boundary fluxes of the orbital rotation and the spin are derived, for the first time, to elucidate their boundary creation mechanisms, respectively. Then, these new findings are successfully extended to the study of the boundary enstrophy dynamics, as well as the Lyman vorticity dynamics as an alternative interpretation to the boundary vorticity dynamics. Interestingly, it is found that the boundary coupling of the longitudinal and transverse processes is only embodied in the boundary spin flux, which is definitely not responsible for the generation of the boundary orbital-rotation flux. In addition, the boundary fluxes of enstrophy are directly associated with the boundary source of the second principal invariant of the velocity gradient tensor, and the two quadratic forms representing the spin-geometry interaction. The present exposition provides a new perspective to the vorticity dynamics on boundaries, which could be valuable in clarifying the formation mechanisms of near-wall coherent structures and flow noise at the fundamental level.

Background: Fairness testing for deep learning systems has been becoming increasingly important. However, much work assumes perfect context and conditions from the other parts: well-tuned hyperparameters for accuracy; rectified bias in data, and mitigated bias in the labeling. Yet, these are often difficult to achieve in practice due to their resource-/labour-intensive nature. Aims: In this paper, we aim to understand how varying contexts affect fairness testing outcomes. Method:We conduct an extensive empirical study, which covers $10,800$ cases, to investigate how contexts can change the fairness testing result at the model level against the existing assumptions. We also study why the outcomes were observed from the lens of correlation/fitness landscape analysis. Results: Our results show that different context types and settings generally lead to a significant impact on the testing, which is mainly caused by the shifts of the fitness landscape under varying contexts. Conclusions: Our findings provide key insights for practitioners to evaluate the test generators and hint at future research directions., Comment: Received by ESEM 24

The recent development of online static map element (a.k.a. HD map) construction algorithms has raised a vast demand for data with ground truth annotations. However, available public datasets currently cannot provide high-quality training data regarding consistency and accuracy. For instance, the manual labelled (low efficiency) nuScenes still contains misalignment and inconsistency between the HD maps and images (e.g., around 8.03 pixels reprojection error on average). To this end, we present CAMAv2: a vision-centric approach for Consistent and Accurate Map Annotation. Without LiDAR inputs, our proposed framework can still generate high-quality 3D annotations of static map elements. Specifically, the annotation can achieve high reprojection accuracy across all surrounding cameras and is spatial-temporal consistent across the whole sequence. We apply our proposed framework to the popular nuScenes dataset to provide efficient and highly accurate annotations. Compared with the original nuScenes static map element, our CAMAv2 annotations achieve lower reprojection errors (e.g., 4.96 vs. 8.03 pixels). Models trained with annotations from CAMAv2 also achieve lower reprojection errors (e.g., 5.62 vs. 8.43 pixels)., Comment: arXiv admin note: text overlap with arXiv:2309.11754

Deep learning models are increasingly deployed on edge Internet of Things (IoT) devices. However, these models typically operate under supervised conditions and fail to recognize unseen classes different from training. To address this, zero-shot learning (ZSL) aims to classify data of unseen classes with the help of semantic information. Foundation models (FMs) trained on web-scale data have shown impressive ZSL capability in natural language processing and visual understanding. However, leveraging FMs' generalized knowledge for zero-shot IoT sensing using signals such as mmWave, IMU, and Wi-Fi has not been fully investigated. In this work, we align the IoT data embeddings with the semantic embeddings generated by an FM's text encoder for zero-shot IoT sensing. To utilize the physics principles governing the generation of IoT sensor signals to derive more effective prompts for semantic embedding extraction, we propose to use cross-attention to combine a learnable soft prompt that is optimized automatically on training data and an auxiliary hard prompt that encodes domain knowledge of the IoT sensing task. To address the problem of IoT embeddings biasing to seen classes due to the lack of unseen class data during training, we propose using data augmentation to synthesize unseen class IoT data for fine-tuning the IoT feature extractor and embedding projector. We evaluate our approach on multiple IoT sensing tasks. Results show that our approach achieves superior open-set detection and generalized zero-shot learning performance compared with various baselines. Our code is available at https://github.com/schrodingho/FM\_ZSL\_IoT.

The recent rise of EEG-based end-to-end deep learning models presents a significant challenge in elucidating how these models process raw EEG signals and generate predictions in the frequency domain. This challenge limits the transparency and credibility of EEG-based end-to-end models, hindering their application in security-sensitive areas. To address this issue, we propose a mask perturbation method to explain the behavior of end-to-end models in the frequency domain. Considering the characteristics of EEG data, we introduce a target alignment loss to mitigate the out-of-distribution problem associated with perturbation operations. Additionally, we develop a perturbation generator to define perturbation generation in the frequency domain. Our explanation method is validated through experiments on multiple representative end-to-end deep learning models in the EEG decoding field, using an established EEG benchmark dataset. The results demonstrate the effectiveness and superiority of our method, and highlight its potential to advance research in EEG-based end-to-end models.

Recent studies have made significant progress in addressing dexterous manipulation problems, particularly in in-hand object reorientation. However, there are few existing works that explore the potential utilization of developed dexterous manipulation controllers for downstream tasks. In this study, we focus on constrained dexterous manipulation for food peeling. Food peeling presents various constraints on the reorientation controller, such as the requirement for the hand to securely hold the object after reorientation for peeling. We propose a simple system for learning a reorientation controller that facilitates the subsequent peeling task. Videos are available at: https://taochenshh.github.io/projects/veg-peeling.

While deep learning, particularly convolutional neural networks (CNNs), has revolutionized remote sensing (RS) change detection (CD), existing approaches often miss crucial features due to neglecting global context and incomplete change learning. Additionally, transformer networks struggle with low-level details. RCTNet addresses these limitations by introducing \textbf{(1)} an early fusion backbone to exploit both spatial and temporal features early on, \textbf{(2)} a Cross-Stage Aggregation (CSA) module for enhanced temporal representation, \textbf{(3)} a Multi-Scale Feature Fusion (MSF) module for enriched feature extraction in the decoder, and \textbf{(4)} an Efficient Self-deciphering Attention (ESA) module utilizing transformers to capture global information and fine-grained details for accurate change detection. Extensive experiments demonstrate RCTNet's clear superiority over traditional RS image CD methods, showing significant improvement and an optimal balance between accuracy and computational cost., Comment: accepted by IEEE Conference on Multimedia Expo

Medical image segmentation has been significantly advanced with the rapid development of deep learning (DL) techniques. Existing DL-based segmentation models are typically discriminative; i.e., they aim to learn a mapping from the input image to segmentation masks. However, these discriminative methods neglect the underlying data distribution and intrinsic class characteristics, suffering from unstable feature space. In this work, we propose to complement discriminative segmentation methods with the knowledge of underlying data distribution from generative models. To that end, we propose a novel hybrid diffusion framework for medical image segmentation, termed HiDiff, which can synergize the strengths of existing discriminative segmentation models and new generative diffusion models. HiDiff comprises two key components: discriminative segmentor and diffusion refiner. First, we utilize any conventional trained segmentation models as discriminative segmentor, which can provide a segmentation mask prior for diffusion refiner. Second, we propose a novel binary Bernoulli diffusion model (BBDM) as the diffusion refiner, which can effectively, efficiently, and interactively refine the segmentation mask by modeling the underlying data distribution. Third, we train the segmentor and BBDM in an alternate-collaborative manner to mutually boost each other. Extensive experimental results on abdomen organ, brain tumor, polyps, and retinal vessels segmentation datasets, covering four widely-used modalities, demonstrate the superior performance of HiDiff over existing medical segmentation algorithms, including the state-of-the-art transformer- and diffusion-based ones. In addition, HiDiff excels at segmenting small objects and generalizing to new datasets. Source codes are made available at https://github.com/takimailto/HiDiff., Comment: Accepted by IEEE Transactions on Medical Imaging 2024

Label noise is ubiquitous in real-world scenarios, posing a practical challenge to supervised models due to its effect in hurting the generalization performance of deep neural networks. Existing methods primarily employ the sample selection paradigm and usually rely on dataset-dependent prior knowledge (\eg, a pre-defined threshold) to cope with label noise, inevitably degrading the adaptivity. Moreover, existing methods tend to neglect the class balance in selecting samples, leading to biased model performance. To this end, we propose a simple yet effective approach named \textbf{SED} to deal with label noise in a \textbf{S}elf-adaptiv\textbf{E} and class-balance\textbf{D} manner. Specifically, we first design a novel sample selection strategy to empower self-adaptivity and class balance when identifying clean and noisy data. A mean-teacher model is then employed to correct labels of noisy samples. Subsequently, we propose a self-adaptive and class-balanced sample re-weighting mechanism to assign different weights to detected noisy samples. Finally, we additionally employ consistency regularization on selected clean samples to improve model generalization performance. Extensive experimental results on synthetic and real-world datasets demonstrate the effectiveness and superiority of our proposed method. The source code has been made available at https://github.com/NUST-Machine-Intelligence-Laboratory/SED., Comment: accepted by the European Conference on Computer Vision (ECCV), 2024

Though adversarial erasing has prevailed in weakly supervised semantic segmentation to help activate integral object regions, existing approaches still suffer from the dilemma of under-activation and over-expansion due to the difficulty in determining when to stop erasing. In this paper, we propose a \textbf{K}nowledge \textbf{T}ransfer with \textbf{S}imulated Inter-Image \textbf{E}rasing (KTSE) approach for weakly supervised semantic segmentation to alleviate the above problem. In contrast to existing erasing-based methods that remove the discriminative part for more object discovery, we propose a simulated inter-image erasing scenario to weaken the original activation by introducing extra object information. Then, object knowledge is transferred from the anchor image to the consequent less activated localization map to strengthen network localization ability. Considering the adopted bidirectional alignment will also weaken the anchor image activation if appropriate constraints are missing, we propose a self-supervised regularization module to maintain the reliable activation in discriminative regions and improve the inter-class object boundary recognition for complex images with multiple categories of objects. In addition, we resort to intra-image erasing and propose a multi-granularity alignment module to gently enlarge the object activation to boost the object knowledge transfer. Extensive experiments and ablation studies on PASCAL VOC 2012 and COCO datasets demonstrate the superiority of our proposed approach. Source codes and models are available at https://github.com/NUST-Machine-Intelligence-Laboratory/KTSE., Comment: accepted by the European Conference on Computer Vision (ECCV), 2024

Unmanned aerial vehicles (UAVs) are vulnerable to interception and attacks when operated remotely without a unified and efficient identity authentication. Meanwhile, the openness of wireless communication environments potentially leads to data leakage and system paralysis. However, conventional authentication schemes in the UAV network are system-centric, failing to adapt to the diversity of UAVs identities and access, resulting in changes in network environments and connection statuses. Additionally, UAVs are not subjected to periodic identity compliance checks once authenticated, leading to difficulties in controlling access anomalies. Therefore, in this work, we consider a zero-trust framework for UAV network authentication, aiming to achieve UAVs identity authentication through the principle of ``never trust and always verify''. We introduce a blockchain-assisted zero-trust authentication scheme, namely BAZAM, designed for multi-UAV wireless networks. In this scheme, UAVs follow a key generation approach using physical unclonable functions (PUFs), and cryptographic technique helps verify registration and access requests of UAVs. The blockchain is applied to store UAVs authentication information in immutable storage. Through thorough security analysis and extensive evaluation, we demonstrate the effectiveness and efficiency of the proposed BAZAM.

This paper studies the problem of pre-training for small models, which is essential for many mobile devices. Current state-of-the-art methods on this problem transfer the representational knowledge of a large network (as a Teacher) into a smaller model (as a Student) using self-supervised distillation, improving the performance of the small model on downstream tasks. However, existing approaches are insufficient in extracting the crucial knowledge that is useful for discerning categories in downstream tasks during the distillation process. In this paper, for the first time, we introduce language guidance to the distillation process and propose a new method named Language-Guided Distillation (LGD) system, which uses category names of the target downstream task to help refine the knowledge transferred between the teacher and student. To this end, we utilize a pre-trained text encoder to extract semantic embeddings from language and construct a textual semantic space called Textual Semantics Bank (TSB). Furthermore, we design a Language-Guided Knowledge Aggregation (LGKA) module to construct the visual semantic space, also named Visual Semantics Bank (VSB). The task-related knowledge is transferred by driving a student encoder to mimic the similarity score distribution inferred by a teacher over TSB and VSB. Compared with other small models obtained by either ImageNet pre-training or self-supervised distillation, experiment results show that the distilled lightweight model using the proposed LGD method presents state-of-the-art performance and is validated on various downstream tasks, including classification, detection, and segmentation. We have made the code available at https://github.com/mZhenz/LGD.

Deep neural networks are proven to be vulnerable to fine-designed adversarial examples, and adversarial defense algorithms draw more and more attention nowadays. Pre-processing based defense is a major strategy, as well as learning robust feature representation has been proven an effective way to boost generalization. However, existing defense works lack considering different depth-level visual features in the training process. In this paper, we first highlight two novel properties of robust features from the feature distribution perspective: 1) \textbf{Diversity}. The robust feature of intra-class samples can maintain appropriate diversity; 2) \textbf{Discriminability}. The robust feature of inter-class samples should ensure adequate separation. We find that state-of-the-art defense methods aim to address both of these mentioned issues well. It motivates us to increase intra-class variance and decrease inter-class discrepancy simultaneously in adversarial training. Specifically, we propose a simple but effective defense based on decoupled visual representation masking. The designed Decoupled Visual Feature Masking (DFM) block can adaptively disentangle visual discriminative features and non-visual features with diverse mask strategies, while the suitable discarding information can disrupt adversarial noise to improve robustness. Our work provides a generic and easy-to-plugin block unit for any former adversarial training algorithm to achieve better protection integrally. Extensive experimental results prove the proposed method can achieve superior performance compared with state-of-the-art defense approaches. The code is publicly available at \href{https://github.com/chenboluo/Adversarial-defense}{https://github.com/chenboluo/Adversarial-defense}., Comment: The code is publicly available

The present work elucidates the boundary behaviors of the velocity gradient tensor ($\bm{A}\equiv\bm{\nabla}\bm{u}$) and its principal invariants ($P,Q,R$) for compressible flow interacting with a stationary rigid wall. Firstly, it is found that the well-known Caswell formula exhibits an inherent physical structure being compatible with the normal-nilpotent decomposition, where both the strain-rate and rotation-rate tensors contain the physical effects from the spin component of the vorticity. Secondly, we derive the kinematic and dynamic forms of the boundary $\bm{A}$-flux from which the known boundary fluxes can be recovered by applying the symmetric-antisymmetric decomposition. Then, we obtain the explicit expression of the boundary $Q$ flux as a result of the competition among the boundary fluxes of squared dilatation, enstrophy and squared strain-rate. Importantly, we emphasize that both the coupling between the spin and surface pressure gradient, and the spin-curvature quadratic interaction, are \textit{not} responsible for the generation of the boundary $Q$ flux, although they contribute to both the boundary fluxes of enstrophy and squared strain-rate. Moreover, we prove that the boundary $R$ flux must vanish on a stationary rigid wall. Finally, the boundary fluxes of the invariants of the strain-rate and rotation-rate tensors are also discussed. It is revealed that the boundary flux of the third invariant of the strain-rate tensor is proportional to the wall-normal derivative of the vortex stretching term, which serves as a source term accounting for the the spatiotemporal evolution rate of the wall-normal enstrophy flux. These theoretical results provide a unified description of boundary vorticity and vortex dynamics, which could be valuable in understanding the formation mechanisms of complex near-wall coherent structures and the boundary sources of flow noise.

Indoor photovoltaics (IPVs) have attracted increasing attention for sustainably powering Internet of Things (IoT) electronics. Sb$_2$S$_3$ is a promising IPV candidate material with a bandgap of ~1.75 eV, which is near the optimal value for indoor energy harvesting. However, the performance of Sb$_2$S$_3$ solar cells is limited by nonradiative recombination, closely associated with the poor-quality absorber films. Additive engineering is an effective strategy to improved the properties of solution-processed films. This work shows that the addition of monoethanolamine (MEA) into the precursor solution allows the nucleation and growth of Sb$_2$S$_3$ films to be controlled, enabling the deposition of high-quality Sb$_2$S$_3$ absorbers with reduced grain boundary density, optimized band positions and increased carrier concentration. Complemented with computations, it is revealed that the incorporation of MEA leads to a more efficient and energetically favorable deposition for enhanced heterogeneous nucleation on the substrate, which increases the grain size and accelerates the deposition rate of Sb$_2$S$_3$ films. Due to suppressed carrier recombination and improved charge-carrier transport in Sb$_2$S$_3$ absorber films, the MEA-modulated Sb$_2$S$_3$ solar cell yields a power conversion efficiency (PCE) of 7.22% under AM1.5G illumination, and an IPV PCE of 17.55% under 1000 lux white light emitting diode (WLED) illumination, which is the highest yet reported for Sb$_2$S$_3$ IPVs. Furthermore, we construct high performance large-area Sb$_2$S$_3$ IPV modules to power IoT wireless sensors, and realize the long-term continuous recording of environmental parameters under WLED illumination in an office. This work highlights the great prospect of Sb$_2$S$_3$ photovoltaics for indoor energy harvesting., Comment: 28 pages, 6 figures

Accurately measuring the three-dimensional thermal conductivity tensor is essential for understanding and engineering the thermal behavior of anisotropic materials. Existing methods often struggle to isolate individual tensor elements, leading to large measurement uncertainties and time-consuming iterative fitting procedures. In this study, we introduce the Beam-Offset Square-Pulsed Source (BO-SPS) method for comprehensive measurements of three-dimensional anisotropic thermal conductivity tensors. This method uses square-pulsed heating and precise temperature rise measurements to achieve high signal-to-noise ratios, even with large beam offsets and low modulation frequencies, enabling the isolation of thermal conductivity tensor elements. We demonstrate and validate the BO-SPS method by measuring X-cut and AT-cut quartz samples. For X-cut quartz, with a known relationship between in-plane and cross-plane thermal conductivities, we can determine the full thermal conductivity tensor and heat capacity simultaneously. For AT-cut quartz, assuming a known heat capacity, we can determine the entire anisotropic thermal conductivity tensor, even with finite off-diagonal terms. Our results yield consistent principal thermal conductivity values for both quartz types, demonstrating the method's reliability and accuracy. This research highlights the BO-SPS method's potential to advance the understanding of thermal behavior in complex materials.

Parameter-efficient fine-tuning (PEFT) has become increasingly important as foundation models continue to grow in both popularity and size. Adapter has been particularly well-received due to their potential for parameter reduction and adaptability across diverse tasks. However, striking a balance between high efficiency and robust generalization across tasks remains a challenge for adapter-based methods. We analyze existing methods and find that: 1) parameter sharing is the key to reducing redundancy; 2) more tunable parameters, dynamic allocation, and block-specific design are keys to improving performance. Unfortunately, no previous work considers all these factors. Inspired by this insight, we introduce a novel framework named Adapter-X. First, a Sharing Mixture of Adapters (SMoA) module is proposed to fulfill token-level dynamic allocation, increased tunable parameters, and inter-block sharing at the same time. Second, some block-specific designs like Prompt Generator (PG) are introduced to further enhance the ability of adaptation. Extensive experiments across 2D image and 3D point cloud modalities demonstrate that Adapter-X represents a significant milestone as it is the first to outperform full fine-tuning in both 2D image and 3D point cloud modalities with significantly fewer parameters, i.e., only 0.20% and 1.88% of original trainable parameters for 2D and 3D classification tasks. Our code will be publicly available.

Diffusion models are widely recognized for generating high-quality and diverse images, but their poor real-time performance has led to numerous acceleration works, primarily focusing on UNet-based structures. With the more successful results achieved by diffusion transformers (DiT), there is still a lack of exploration regarding the impact of DiT structure on generation, as well as the absence of an acceleration framework tailored to the DiT architecture. To tackle these challenges, we conduct an investigation into the correlation between DiT blocks and image generation. Our findings reveal that the front blocks of DiT are associated with the outline of the generated images, while the rear blocks are linked to the details. Based on this insight, we propose an overall training-free inference acceleration framework $\Delta$-DiT: using a designed cache mechanism to accelerate the rear DiT blocks in the early sampling stages and the front DiT blocks in the later stages. Specifically, a DiT-specific cache mechanism called $\Delta$-Cache is proposed, which considers the inputs of the previous sampling image and reduces the bias in the inference. Extensive experiments on PIXART-$\alpha$ and DiT-XL demonstrate that the $\Delta$-DiT can achieve a $1.6\times$ speedup on the 20-step generation and even improves performance in most cases. In the scenario of 4-step consistent model generation and the more challenging $1.12\times$ acceleration, our method significantly outperforms existing methods. Our code will be publicly available., Comment: 12 pages, 6 figures, 6 tables

Data assimilation is a vital component in modern global medium-range weather forecasting systems to obtain the best estimation of the atmospheric state by combining the short-term forecast and observations. Recently, AI-based data assimilation approaches have attracted increasing attention for their significant advantages over traditional techniques in terms of computational consumption. However, existing AI-based data assimilation methods can only handle observations with a specific resolution, lacking the compatibility and generalization ability to assimilate observations with other resolutions. Considering that complex real-world observations often have different resolutions, we propose the \textit{\textbf{Fourier Neural Processes}} (FNP) for \textit{arbitrary-resolution data assimilation} in this paper. Leveraging the efficiency of the designed modules and flexible structure of neural processes, FNP achieves state-of-the-art results in assimilating observations with varying resolutions, and also exhibits increasing advantages over the counterparts as the resolution and the amount of observations increase. Moreover, our FNP trained on a fixed resolution can directly handle the assimilation of observations with out-of-distribution resolutions and the observational information reconstruction task without additional fine-tuning, demonstrating its excellent generalization ability across data resolutions as well as across tasks.

Deep reinforcement learning (RL) algorithms typically parameterize the policy as a deep network that outputs either a deterministic action or a stochastic one modeled as a Gaussian distribution, hence restricting learning to a single behavioral mode. Meanwhile, diffusion models emerged as a powerful framework for multimodal learning. However, the use of diffusion policies in online RL is hindered by the intractability of policy likelihood approximation, as well as the greedy objective of RL methods that can easily skew the policy to a single mode. This paper presents Deep Diffusion Policy Gradient (DDiffPG), a novel actor-critic algorithm that learns from scratch multimodal policies parameterized as diffusion models while discovering and maintaining versatile behaviors. DDiffPG explores and discovers multiple modes through off-the-shelf unsupervised clustering combined with novelty-based intrinsic motivation. DDiffPG forms a multimodal training batch and utilizes mode-specific Q-learning to mitigate the inherent greediness of the RL objective, ensuring the improvement of the diffusion policy across all modes. Our approach further allows the policy to be conditioned on mode-specific embeddings to explicitly control the learned modes. Empirical studies validate DDiffPG's capability to master multimodal behaviors in complex, high-dimensional continuous control tasks with sparse rewards, also showcasing proof-of-concept dynamic online replanning when navigating mazes with unseen obstacles.

State-of-the-art techniques like dual-frequency Time-Domain Thermoreflectance (TDTR) and Frequency-Domain Thermoreflectance (FDTR) offer superb capability for simultaneous measurements of thermal conductivity and heat capacity with a spatial resolution on the order of 10 {\mu}m. However, their applicability is limited to highly conductive materials with an in-plane thermal conductivity exceeding 10 W/(m*K). In this paper, we introduce the Square-Pulsed Source (SPS) technique, offering a novel approach to concurrently measure thermal conductivity and heat capacity with a 10 {\mu}m spatial resolution, while significantly extending the measurable thermal conductivity range to an unprecedented low of 0.1 W/(m*K), offering enhanced versatility. To demonstrate and validate its efficacy, we conducted measurements on various standard materials--PMMA, silica, sapphire, silicon, and diamond--spanning a wide thermal conductivity range from 0.1 to 2000 W/(m*K). The obtained results exhibit remarkable agreement with literature values, with a typical measurement uncertainty of less than 10% across the entire thermal conductivity range. By providing a unique capability to characterize both highly and lowly conductive materials with micron-scale spatial resolution, the SPS method opens new avenues for understanding and engineering thermal properties across diverse applications.

The polygon mesh representation of 3D data exhibits great flexibility, fast rendering speed, and storage efficiency, which is widely preferred in various applications. However, given its unstructured graph representation, the direct generation of high-fidelity 3D meshes is challenging. Fortunately, with a pre-defined ordering strategy, 3D meshes can be represented as sequences, and the generation process can be seamlessly treated as an auto-regressive problem. In this paper, we validate the Neural Coordinate Field (NeurCF), an explicit coordinate representation with implicit neural embeddings, is a simple-yet-effective representation for large-scale sequential mesh modeling. After that, we present MeshXL, a family of generative pre-trained auto-regressive models, which addresses the process of 3D mesh generation with modern large language model approaches. Extensive experiments show that MeshXL is able to generate high-quality 3D meshes, and can also serve as foundation models for various down-stream applications.

The success of pretrain-finetune paradigm brings about the release of numerous model weights. In this case, merging models finetuned on different tasks to enable a single model with multi-task capabilities is gaining increasing attention for its practicability. Existing model merging methods usually suffer from (1) significant performance degradation or (2) requiring tuning by additional data or training. In this paper, we rethink and analyze the existing model merging paradigm. We discover that using a single model's weights can hardly simulate all the models' performance. To tackle this issue, we propose Elect, Mask & Rescale-Merging (EMR-Merging). We first (a) elect a unified model from all the model weights and then (b) generate extremely lightweight task-specific modulators, including masks and rescalers, to align the direction and magnitude between the unified model and each specific model, respectively. EMR-Merging is tuning-free, thus requiring no data availability or any additional training while showing impressive performance. We find that EMR-Merging shows outstanding performance compared to existing merging methods under different classical and newly-established settings, including merging different numbers of vision models (up to 30), NLP models, PEFT models, and multi-modal models., Comment: NeurIPS 2024

The half-size prototype of the multi wire drift chamber (MWDC) for the cooling storage ring (CSR) external-target experiment (CEE) was assembled and tested in 350 MeV/u Kr+Fe reactions on the heavy ion research facility in Lanzhou (HIRFL). The prototype consists of 6 sense layers, where the sense wires are stretched in three directions X, U and V, meeting $0^\circ$, $30^\circ$ and $-30^\circ$ with respect to the vertical axis, respectively. The sensitive area of the prototype is $76 {\rm cm} \times 76 {\rm cm}$. The amplified and shaped signals from the anode wires are digitized in a serial capacity array. Being operated with 1500 V high voltage on the anode wires, the efficiency for each layer is beyond 95\%. The tracking residual is about $301 \pm 2 \rm \mu m$. The performance meets the requirements of CEE.

Despite their impressive generative performance, latent diffusion model-based virtual try-on (VTON) methods lack faithfulness to crucial details of the clothes, such as style, pattern, and text. To alleviate these issues caused by the diffusion stochastic nature and latent supervision, we propose a novel Faithful Latent Diffusion Model for VTON, termed FLDM-VTON. FLDM-VTON improves the conventional latent diffusion process in three major aspects. First, we propose incorporating warped clothes as both the starting point and local condition, supplying the model with faithful clothes priors. Second, we introduce a novel clothes flattening network to constrain generated try-on images, providing clothes-consistent faithful supervision. Third, we devise a clothes-posterior sampling for faithful inference, further enhancing the model performance over conventional clothes-agnostic Gaussian sampling. Extensive experimental results on the benchmark VITON-HD and Dress Code datasets demonstrate that our FLDM-VTON outperforms state-of-the-art baselines and is able to generate photo-realistic try-on images with faithful clothing details., Comment: Accepted by IJCAI 2024

In the Massive Open Online Courses (MOOC) learning scenario, the semantic information of instructional videos has a crucial impact on learners' emotional state. Learners mainly acquire knowledge by watching instructional videos, and the semantic information in the videos directly affects learners' emotional states. However, few studies have paid attention to the potential influence of the semantic information of instructional videos on learners' emotional states. To deeply explore the impact of video semantic information on learners' emotions, this paper innovatively proposes a multimodal emotion recognition method by fusing video semantic information and physiological signals. We generate video descriptions through a pre-trained large language model (LLM) to obtain high-level semantic information about instructional videos. Using the cross-attention mechanism for modal interaction, the semantic information is fused with the eye movement and PhotoPlethysmoGraphy (PPG) signals to obtain the features containing the critical information of the three modes. The accurate recognition of learners' emotional states is realized through the emotion classifier. The experimental results show that our method has significantly improved emotion recognition performance, providing a new perspective and efficient method for emotion recognition research in MOOC learning scenarios. The method proposed in this paper not only contributes to a deeper understanding of the impact of instructional videos on learners' emotional states but also provides a beneficial reference for future research on emotion recognition in MOOC learning scenarios.

When tuning software configuration for better performance (e.g., latency or throughput), an important issue that many optimizers face is the presence of local optimum traps, compounded by a highly rugged configuration landscape and expensive measurements. To mitigate these issues, a recent effort has shifted to focus on the level of optimization model (called meta multi-objectivization or MMO) instead of designing better optimizers as in traditional methods. This is done by using an auxiliary performance objective, together with the target performance objective, to help the search jump out of local optima. While effective, MMO needs a fixed weight to balance the two objectives-a parameter that has been found to be crucial as there is a large deviation of the performance between the best and the other settings. However, given the variety of configurable software systems, the "sweet spot" of the weight can vary dramatically in different cases and it is not possible to find the right setting without time-consuming trial and error. In this paper, we seek to overcome this significant shortcoming of MMO by proposing a weight adaptation method, dubbed AdMMO. Our key idea is to adaptively adjust the weight at the right time during tuning, such that a good proportion of the nondominated configurations can be maintained. Moreover, we design a partial duplicate retention mechanism to handle the issue of too many duplicate configurations without losing the rich information provided by the "good" duplicates. Experiments on several real-world systems, objectives, and budgets show that, for 71% of the cases, AdMMO is considerably superior to MMO and a wide range of state-of-the-art optimizers while achieving generally better efficiency with the best speedup between 2.2x and 20x., Comment: This paper has been accepted at ACM FSE'24

Operators experience complicated physiological and psychological states when exposed to extreme heat stress, which can impair cognitive function and decrease performance significantly, ultimately leading to severe secondary disasters. Therefore, there is an urgent need for a feasible technique to identify their abnormal states to enhance the reliability of human-cybernetics systems. With the advancement of deep learning in physiological modeling, a model for evaluating operators' performance driven by electrocardiogram (ECG) and functional near-infrared spectroscopy (fNIRS) was proposed, demonstrating high ecological validity. The model fused a convolutional neural network (CNN) backbone and a graph attention network (GAT) backbone to extract discriminative features from ECG time-frequency spectrums and fNIRS prefrontal cortex (PFC) network respectively with deeper neuroscience domain knowledge, and eventually achieved 0.90 AUC. Results supported that handcrafted features extracted by specialized neuroscience methods can alleviate overfitting. Inspired by the small-world nature of the brain network, the fNIRS PFC network was organized as an undirected graph and embedded by GAT. It is proven to perform better in information aggregation and delivery compared to a simple non-linear transformation. The model provides a potential neuroergonomics application for evaluating the human state in vital human-cybernetics systems under industry 5.0 scenarios.

Coherent dynamics of interacting quantum particles plays a central role in the study of strongly correlated quantum matter and the pursuit of quantum information processors. Here, we present the state-space of interacting Rydberg atoms as a synthetic landscape on which to control and observe coherent and correlated dynamics. With full control of the coupling strengths and energy offsets between the pairs of sites in a nine-site synthetic lattice, we realize quantum walks, Bloch oscillations, and dynamics in an Escher-type "continuous staircase". In the interacting regime, we observe correlated quantum walks, Bloch oscillations, and confinement of particle pairs. Additionally, we simultaneously tilt our lattice both up and down to achieve coherent pair oscillations. When combined with a few straightforward upgrades, this work establishes synthetic Rydberg lattices of interacting atom arrays as a promising platform for programmable quantum many-body dynamics with access to features that are difficult to realize in real-space lattices., Comment: 5 pages, 4 figures

Flat bands play a central role in hosting emergent states of matter in many condensed matter systems, from the nascent insulating states of twisted bilayer graphene to the fractionalized excitations found in frustrated magnets and quantum Hall materials. Here, we report on the experimental realization of highly tunable flat-band models populated by strongly interacting Rydberg atoms. Using the approach of synthetic dimensions, we engineer a flat-band rhombic lattice with twisted boundaries, and through nonequilibrium dynamics we explore the control of Aharonov--Bohm (AB) caging via a tunable $U(1)$ gauge field. Through microscopic measurements of Rydberg pairs, we explore the interaction-driven breakdown of AB caging in the limit of strong dipolar interactions that mix the lattice bands. In the limit of weak interactions, where caging remains intact, we observe an effective magnetism that arises due to the interaction-driven mixing of degenerate flat-band states. These observations of strongly correlated flat-band dynamics open the door to explorations of new emergent phenomena in synthetic quantum materials., Comment: 7 pages, 4 figures

Recent Vision Transformer Compression (VTC) works mainly follow a two-stage scheme, where the importance score of each model unit is first evaluated or preset in each submodule, followed by the sparsity score evaluation according to the target sparsity constraint. Such a separate evaluation process induces the gap between importance and sparsity score distributions, thus causing high search costs for VTC. In this work, for the first time, we investigate how to integrate the evaluations of importance and sparsity scores into a single stage, searching the optimal subnets in an efficient manner. Specifically, we present OFB, a cost-efficient approach that simultaneously evaluates both importance and sparsity scores, termed Once for Both (OFB), for VTC. First, a bi-mask scheme is developed by entangling the importance score and the differentiable sparsity score to jointly determine the pruning potential (prunability) of each unit. Such a bi-mask search strategy is further used together with a proposed adaptive one-hot loss to realize the progressive-and-efficient search for the most important subnet. Finally, Progressive Masked Image Modeling (PMIM) is proposed to regularize the feature space to be more representative during the search process, which may be degraded by the dimension reduction. Extensive experiments demonstrate that OFB can achieve superior compression performance over state-of-the-art searching-based and pruning-based methods under various Vision Transformer architectures, meanwhile promoting search efficiency significantly, e.g., costing one GPU search day for the compression of DeiT-S on ImageNet-1K., Comment: Accepted by CVPR 2024. Our code will be available at www.github.com/HankYe/Once-for-Both

The ability to model the underlying dynamics of visual scenes and reason about the future is central to human intelligence. Many attempts have been made to empower intelligent systems with such physical understanding and prediction abilities. However, most existing methods focus on pixel-to-pixel prediction, which suffers from heavy computational costs while lacking a deep understanding of the physical dynamics behind videos. Recently, object-centric prediction methods have emerged and attracted increasing interest. Inspired by it, this paper proposes an unsupervised object-centric prediction model that makes future predictions by learning visual dynamics between objects. Our model consists of two modules, perceptual, and dynamic module. The perceptual module is utilized to decompose images into several objects and synthesize images with a set of object-centric representations. The dynamic module fuses contextual information, takes environment-object and object-object interaction into account, and predicts the future trajectory of objects. Extensive experiments are conducted to validate the effectiveness of the proposed method. Both quantitative and qualitative experimental results demonstrate that our model generates higher visual quality and more physically reliable predictions compared to the state-of-the-art methods., Comment: 13 pages, 10 figures

In remote estimation of cyber-physical systems (CPSs), sensor measurements transmitted through network may be attacked by adversaries, leading to leakage risk of privacy (e.g., the system state), and/or failure of the remote estimator. To deal with this problem, a chaotic masking protocol is proposed in this paper to secure the sensor measurements transmission. In detail, at the plant side, a chaotic dynamic system is deployed to encode the sensor measurement, and at the estimator side, an estimator estimates both states of the physical plant and the chaotic system. With this protocol, no additional secure communication links is needed for synchronization, and the masking effect can be perfectly removed when the estimator is in steady state. Furthermore, this masking protocol can deal with multiple types of attacks, i.e., eavesdropping attack, replay attack, and stealthy false data injection attack., Comment: 8 pages, 7 figures

Sparsification-based pruning has been an important category in model compression. Existing methods commonly set sparsity-inducing penalty terms to suppress the importance of dropped weights, which is regarded as the suppressed sparsification paradigm. However, this paradigm inactivates the dropped parts of networks causing capacity damage before pruning, thereby leading to performance degradation. To alleviate this issue, we first study and reveal the relative sparsity effect in emerging stimulative training and then propose a structured pruning framework, named STP, based on an enhanced sparsification paradigm which maintains the magnitude of dropped weights and enhances the expressivity of kept weights by self-distillation. Besides, to find an optimal architecture for the pruned network, we propose a multi-dimension architecture space and a knowledge distillation-guided exploration strategy. To reduce the huge capacity gap of distillation, we propose a subnet mutating expansion technique. Extensive experiments on various benchmarks indicate the effectiveness of STP. Specifically, without fine-tuning, our method consistently achieves superior performance at different budgets, especially under extremely aggressive pruning scenarios, e.g., remaining 95.11% Top-1 accuracy (72.43% in 76.15%) while reducing 85% FLOPs for ResNet-50 on ImageNet. Codes will be released soon., Comment: 26 pages

While various deep learning methods were proposed for low-dose computed tomography (CT) denoising, they often suffer from over-smoothing, blurring, and lack of explainability. To alleviate these issues, we propose a plug-and-play Language-Engaged Dual-space Alignment loss (LEDA) to optimize low-dose CT denoising models. Our idea is to leverage large language models (LLMs) to align denoised CT and normal dose CT images in both the continuous perceptual space and discrete semantic space, which is the first LLM-based scheme for low-dose CT denoising. LEDA involves two steps: the first is to pretrain an LLM-guided CT autoencoder, which can encode a CT image into continuous high-level features and quantize them into a token space to produce semantic tokens derived from the LLM's vocabulary; and the second is to minimize the discrepancy between the denoised CT images and normal dose CT in terms of both encoded high-level features and quantized token embeddings derived by the LLM-guided CT autoencoder. Extensive experimental results on two public LDCT denoising datasets demonstrate that our LEDA can enhance existing denoising models in terms of quantitative metrics and qualitative evaluation, and also provide explainability through language-level image understanding. Source code is available at https://github.com/hao1635/LEDA., Comment: 11 pages, 6 figures

Within the domain of medical analysis, extensive research has explored the potential of mutual learning between Masked Autoencoders(MAEs) and multimodal data. However, the impact of MAEs on intermodality remains a key challenge. We introduce MedFLIP, a Fast Language-Image Pre-training method for Medical analysis. We explore MAEs for zero-shot learning with crossed domains, which enhances the model's ability to learn from limited data, a common scenario in medical diagnostics. We verify that masking an image does not affect inter-modal learning. Furthermore, we propose the SVD loss to enhance the representation learning for characteristics of medical images, aiming to improve classification accuracy by leveraging the structural intricacies of such data. Our theory posits that masking encourages semantic preservation, robust feature extraction, regularization, domain adaptation, and invariance learning. Lastly, we validate using language will improve the zero-shot performance for the medical image analysis. MedFLIP's scaling of the masking process marks an advancement in the field, offering a pathway to rapid and precise medical image analysis without the traditional computational bottlenecks. Through experiments and validation, MedFLIP demonstrates efficient performance improvements, helps for future research and application in medical diagnostics.

EEG-based Emotion recognition holds significant promise for applications in human-computer interaction, medicine, and neuroscience. While deep learning has shown potential in this field, current approaches usually rely on large-scale high-quality labeled datasets, limiting the performance of deep learning. Self-supervised learning offers a solution by automatically generating labels, but its inter-subject generalizability remains under-explored. For this reason, our interest lies in offering a self-supervised learning paradigm with better inter-subject generalizability. Inspired by recent efforts in combining low-level and high-level tasks in deep learning, we propose a cascaded self-supervised architecture for EEG emotion recognition. Then, we introduce a low-level task, time-to-frequency reconstruction (TFR). This task leverages the inherent time-frequency relationship in EEG signals. Our architecture integrates it with the high-level contrastive learning modules, performing self-supervised learning for EEG-based emotion recognition. Experiment on DEAP and DREAMER datasets demonstrates superior performance of our method over similar works. The outcome results also highlight the indispensability of the TFR task and the robustness of our method to label scarcity, validating the effectiveness of the proposed method.

Imitation learning methods need significant human supervision to learn policies robust to changes in object poses, physical disturbances, and visual distractors. Reinforcement learning, on the other hand, can explore the environment autonomously to learn robust behaviors but may require impractical amounts of unsafe real-world data collection. To learn performant, robust policies without the burden of unsafe real-world data collection or extensive human supervision, we propose RialTo, a system for robustifying real-world imitation learning policies via reinforcement learning in "digital twin" simulation environments constructed on the fly from small amounts of real-world data. To enable this real-to-sim-to-real pipeline, RialTo proposes an easy-to-use interface for quickly scanning and constructing digital twins of real-world environments. We also introduce a novel "inverse distillation" procedure for bringing real-world demonstrations into simulated environments for efficient fine-tuning, with minimal human intervention and engineering required. We evaluate RialTo across a variety of robotic manipulation problems in the real world, such as robustly stacking dishes on a rack, placing books on a shelf, and six other tasks. RialTo increases (over 67%) in policy robustness without requiring extensive human data collection. Project website and videos at https://real-to-sim-to-real.github.io/RialTo/, Comment: Project page: https://real-to-sim-to-real.github.io/RialTo/

Performance is arguably the most crucial attribute that reflects the quality of a configurable software system. However, given the increasing scale and complexity of modern software, modeling and predicting how various configurations can impact performance becomes one of the major challenges in software maintenance. As such, performance is often modeled without having a thorough knowledge of the software system, but relying mainly on data, which fits precisely with the purpose of deep learning. In this paper, we conduct a comprehensive review exclusively on the topic of deep learning for performance learning of configurable software, covering 1,206 searched papers spanning six indexing services, based on which 99 primary papers were extracted and analyzed. Our results outline key statistics, taxonomy, strengths, weaknesses, and optimal usage scenarios for techniques related to the preparation of configuration data, the construction of deep learning performance models, the evaluation of these models, and their utilization in various software configuration-related tasks.We also identify the good practices and potentially problematic phenomena from the studies surveyed, together with a comprehensive summary of actionable suggestions and insights into future opportunities within the field. To promote open science, all the raw results of this survey can be accessed at our repository: https://github.com/ideas-labo/DCPL-SLR., Comment: Accepted by the TOSEM survey track

Vision-Language Transformers (VLTs) have shown great success recently, but are meanwhile accompanied by heavy computation costs, where a major reason can be attributed to the large number of visual and language tokens. Existing token pruning research for compressing VLTs mainly follows a single-modality-based scheme yet ignores the critical role of aligning different modalities for guiding the token pruning process, causing the important tokens for one modality to be falsely pruned in another modality branch. Meanwhile, existing VLT pruning works also lack the flexibility to dynamically compress each layer based on different input samples. To this end, we propose a novel framework named Multimodal Alignment-Guided Dynamic Token Pruning (MADTP) for accelerating various VLTs. Specifically, we first introduce a well-designed Multi-modality Alignment Guidance (MAG) module that can align features of the same semantic concept from different modalities, to ensure the pruned tokens are less important for all modalities. We further design a novel Dynamic Token Pruning (DTP) module, which can adaptively adjust the token compression ratio in each layer based on different input instances. Extensive experiments on various benchmarks demonstrate that MADTP significantly reduces the computational complexity of kinds of multimodal models while preserving competitive performance. Notably, when applied to the BLIP model in the NLVR2 dataset, MADTP can reduce the GFLOPs by 80% with less than 4% performance degradation., Comment: 19 pages, 9 figures, Published in CVPR2024