Search

The Mantel-Haenszel (MH) risk difference estimator, commonly used in randomized clinical trials for binary outcomes, calculates a weighted average of stratum-specific risk difference estimators. Traditionally, this method requires the stringent assumption that risk differences are homogeneous across strata, also known as the common risk difference assumption. In our article, we relax this assumption and adopt a modern perspective, viewing the MH risk difference estimator as an approach for covariate adjustment in randomized clinical trials, distinguishing its use from that in meta-analysis and observational studies. We demonstrate that the MH risk difference estimator consistently estimates the average treatment effect within a standard super-population framework, which is often the primary interest in randomized clinical trials, in addition to estimating a weighted average of stratum-specific risk difference. We rigorously study its properties under both the large-stratum and sparse-stratum asymptotic regimes. Furthermore, for either estimand, we propose a unified robust variance estimator that improves over the popular variance estimators by Greenland and Robins (1985) and Sato et al. (1989) and has provable consistency across both asymptotic regimes, regardless of assuming common risk differences. Extensions of our theoretical results also provide new insights into the Cochran-Mantel-Haenszel test and the post-stratification estimator. Our findings are thoroughly validated through simulations and a clinical trial example.

In this work, we improved the analysis of the running time of SparseGPT [Frantar, Alistarh ICML 2023] from $O(d^{3})$ to $O(d^{\omega} + d^{2+a+o(1)} + d^{1+\omega(1,1,a)-a})$ for any $a \in [0, 1]$, where $\omega$ is the exponent of matrix multiplication. In particular, for the current $\omega \approx 2.371$ [Alman, Duan, Williams, Xu, Xu, Zhou 2024], our running times boil down to $O(d^{2.53})$. This running time is due to the analysis of the lazy update behavior in iterative maintenance problems, such as [Deng, Song, Weinstein 2022, Brand, Song, Zhou ICML 2024].

Text-driven video editing has recently experienced rapid development. Despite this, evaluating edited videos remains a considerable challenge. Current metrics tend to fail to align with human perceptions, and effective quantitative metrics for video editing are still notably absent. To address this, we introduce E-Bench, a benchmark suite tailored to the assessment of text-driven video editing. This suite includes E-Bench DB, a video quality assessment (VQA) database for video editing. E-Bench DB encompasses a diverse set of source videos featuring various motions and subjects, along with multiple distinct editing prompts, editing results from 8 different models, and the corresponding Mean Opinion Scores (MOS) from 24 human annotators. Based on E-Bench DB, we further propose E-Bench QA, a quantitative human-aligned measurement for the text-driven video editing task. In addition to the aesthetic, distortion, and other visual quality indicators that traditional VQA methods emphasize, E-Bench QA focuses on the text-video alignment and the relevance modeling between source and edited videos. It proposes a new assessment network for video editing that attains superior performance in alignment with human preferences. To the best of our knowledge, E-Bench introduces the first quality assessment dataset for video editing and an effective subjective-aligned quantitative metric for this domain. All data and code will be publicly available at https://github.com/littlespray/E-Bench.

In recent years, the field of visual tracking has made significant progress with the application of large-scale training datasets. These datasets have supported the development of sophisticated algorithms, enhancing the accuracy and stability of visual object tracking. However, most research has primarily focused on favorable illumination circumstances, neglecting the challenges of tracking in low-ligh environments. In low-light scenes, lighting may change dramatically, targets may lack distinct texture features, and in some scenarios, targets may not be directly observable. These factors can lead to a severe decline in tracking performance. To address this issue, we introduce LLOT, a benchmark specifically designed for Low-Light Object Tracking. LLOT comprises 269 challenging sequences with a total of over 132K frames, each carefully annotated with bounding boxes. This specially designed dataset aims to promote innovation and advancement in object tracking techniques for low-light conditions, addressing challenges not adequately covered by existing benchmarks. To assess the performance of existing methods on LLOT, we conducted extensive tests on 39 state-of-the-art tracking algorithms. The results highlight a considerable gap in low-light tracking performance. In response, we propose H-DCPT, a novel tracker that incorporates historical and darkness clue prompts to set a stronger baseline. H-DCPT outperformed all 39 evaluated methods in our experiments, demonstrating significant improvements. We hope that our benchmark and H-DCPT will stimulate the development of novel and accurate methods for tracking objects in low-light conditions. The LLOT and code are available at https://github.com/OpenCodeGithub/H-DCPT.

Recently, the LHCb Collaboration has measured two decay processes $B^0\to\bar{D}^0D_s^+\pi^-$ and $B^+\to D^-D_s^+\pi^+$ related to isospin symmetry, where two new open-flavor tetraquark states $T_{c\bar{s}}(2900)^0$ and $T_{c\bar{s}}(2900)^{++}$ that belong to an isospin triplet were observed in the $D_s^+\pi^-$ and $D_s^+\pi^+$ invariant mass distributions. In this work, we have investigated the validity of the process $\bar{B}_s^0\to K^0D^0\pi^0$ as the promising process to confirm the existence of $T_{c\bar{s}}(2900)^0$ resonance. Taking into account the tetraquark state $T_{c\bar{s}}(2900)$, as well as intermediate resonances $K^*(892)$, $K_0^*(1430)$, and $K_2^*(1430)$, it has been shown that a clear peak of the open-flavor tetraquark $T_{c\bar{s}}(2900)$ appears in the $K^0D^0$ invariant mass distribution of the process $\bar{B}_s^0\to K^0D^0\pi^0$, which could be tested by future experiments., Comment: 10 pages, 10 figures, comments are welcome!

The global positioning system (GPS) has become an indispensable navigation method for field operations with unmanned surface vehicles (USVs) in marine environments. However, GPS may not always be available outdoors because it is vulnerable to natural interference and malicious jamming attacks. Thus, an alternative navigation system is required when the use of GPS is restricted or prohibited. To this end, we present a novel method that utilizes an Unmanned Aerial Vehicle (UAV) to assist in localizing USVs in GNSS-restricted marine environments. In our approach, the UAV flies along the shoreline at a consistent altitude, continuously tracking and detecting the USV using a deep learning-based approach on camera images. Subsequently, triangulation techniques are applied to estimate the USV's position relative to the UAV, utilizing geometric information and datalink range from the UAV. We propose adjusting the UAV's camera angle based on the pixel error between the USV and the image center throughout the localization process to enhance accuracy. Additionally, visual measurements are integrated into an Extended Kalman Filter (EKF) for robust state estimation. To validate our proposed method, we utilize a USV equipped with onboard sensors and a UAV equipped with a camera. A heterogeneous robotic interface is established to facilitate communication between the USV and UAV. We demonstrate the efficacy of our approach through a series of experiments conducted during the ``Muhammad Bin Zayed International Robotic Challenge (MBZIRC-2024)'' in real marine environments, incorporating noisy measurements and ocean disturbances. The successful outcomes indicate the potential of our method to complement GPS for USV navigation.

Effectively summarizing dense 3D point cloud data and extracting motion information of moving objects (moving object segmentation, MOS) is crucial to autonomous driving and robotics applications. How to effectively utilize motion and semantic features and avoid information loss during 3D-to-2D projection is still a key challenge. In this paper, we propose a novel multi-view MOS model (MV-MOS) by fusing motion-semantic features from different 2D representations of point clouds. To effectively exploit complementary information, the motion branches of the proposed model combines motion features from both bird's eye view (BEV) and range view (RV) representations. In addition, a semantic branch is introduced to provide supplementary semantic features of moving objects. Finally, a Mamba module is utilized to fuse the semantic features with motion features and provide effective guidance for the motion branches. We validated the effectiveness of the proposed multi-branch fusion MOS framework via comprehensive experiments, and our proposed model outperforms existing state-of-the-art models on the SemanticKITTI benchmark., Comment: 7 pages, 4 figures

Let $\epsilon_{1},\ldots,\epsilon_{n}$ be a sequence of independent Rademacher random variables. We prove that there is a constant $c>0$ such that for any unit vectors $v_1,\ldots,v_n\in \mathbb{R}^2$, $$\Pr\left[||\epsilon_1 v_1+\ldots+\epsilon_n v_n||_2 \leq \sqrt{2}\right]\geq \frac{c}{n}.$$ This resolves the only remaining conjecture from the seminal paper of Erd\H{o}s on the Littlewood--Offord problem, and it is sharp both in the sense that the constant $\sqrt{2}$ cannot be reduced and that the magnitude $n^{-1}$ is best possible. We also prove polynomial bounds for the analogous problem in higher dimensions., Comment: 17 pages

Large language models (LLMs) are trained on extensive text corpora, which inevitably include biased information. Although techniques such as Affective Alignment can mitigate some negative impacts of these biases, existing prompt-based attack methods can still extract these biases from the model's weights. Moreover, these biases frequently appear subtly when LLMs are prompted to perform identical tasks across different demographic groups, thereby camouflaging their presence. To address this issue, we have formally defined the implicit bias problem and developed an innovative framework for bias removal based on Bayesian theory, Bayesian-Theory based Bias Removal (BTBR). BTBR employs likelihood ratio screening to pinpoint data entries within publicly accessible biased datasets that represent biases inadvertently incorporated during the LLM training phase. It then automatically constructs relevant knowledge triples and expunges bias information from LLMs using model editing techniques. Through extensive experimentation, we have confirmed the presence of the implicit bias problem in LLMs and demonstrated the effectiveness of our BTBR approach.

Sequential recommendation systems predict a user's next item of interest by analyzing past interactions, aligning recommendations with individual preferences. Leveraging the strengths of Large Language Models (LLMs) in knowledge comprehension and reasoning, recent approaches have applied LLMs to sequential recommendation through language generation paradigms. These methods convert user behavior sequences into prompts for LLM fine-tuning, utilizing Low-Rank Adaptation (LoRA) modules to refine recommendations. However, the uniform application of LoRA across diverse user behaviors sometimes fails to capture individual variability, leading to suboptimal performance and negative transfer between disparate sequences. To address these challenges, we propose Instance-wise LoRA (iLoRA), integrating LoRA with the Mixture of Experts (MoE) framework. iLoRA creates a diverse array of experts, each capturing specific aspects of user preferences, and introduces a sequence representation guided gate function. This gate function processes historical interaction sequences to generate enriched representations, guiding the gating network to output customized expert participation weights. This tailored approach mitigates negative transfer and dynamically adjusts to diverse behavior patterns. Extensive experiments on three benchmark datasets demonstrate the effectiveness of iLoRA, highlighting its superior performance compared to existing methods in capturing user-specific preferences and improving recommendation accuracy.

An ideal dialogue system requires continuous skill acquisition and adaptation to new tasks while retaining prior knowledge. Dialogue State Tracking (DST), vital in these systems, often involves learning new services and confronting catastrophic forgetting, along with a critical capability loss termed the "Value Selection Quandary." To address these challenges, we introduce the Reason-of-Select (RoS) distillation method by enhancing smaller models with a novel 'meta-reasoning' capability. Meta-reasoning employs an enhanced multi-domain perspective, combining fragments of meta-knowledge from domain-specific dialogues during continual learning. This transcends traditional single-perspective reasoning. The domain bootstrapping process enhances the model's ability to dissect intricate dialogues from multiple possible values. Its domain-agnostic property aligns data distribution across different domains, effectively mitigating forgetting. Additionally, two novel improvements, "multi-value resolution" strategy and Semantic Contrastive Reasoning Selection method, significantly enhance RoS by generating DST-specific selection chains and mitigating hallucinations in teachers' reasoning, ensuring effective and reliable knowledge transfer. Extensive experiments validate the exceptional performance and robust generalization capabilities of our method. The source code is provided for reproducibility., Comment: Accepted to ACL 2024 Findings

Unfolded proximal neural networks (PNNs) form a family of methods that combines deep learning and proximal optimization approaches. They consist in designing a neural network for a specific task by unrolling a proximal algorithm for a fixed number of iterations, where linearities can be learned from prior training procedure. PNNs have shown to be more robust than traditional deep learning approaches while reaching at least as good performances, in particular in computational imaging. However, training PNNs still depends on the efficiency of available training algorithms. In this work, we propose a lifted training formulation based on Bregman distances for unfolded PNNs. Leveraging the deterministic mini-batch block-coordinate forward-backward method, we design a bespoke computational strategy beyond traditional back-propagation methods for solving the resulting learning problem efficiently. We assess the behaviour of the proposed training approach for PNNs through numerical simulations on image denoising, considering a denoising PNN whose structure is based on dual proximal-gradient iterations., Comment: 2024 IEEE International Workshop on Machine Learning for Signal Processing, Sept. 22--25, 2024, London, UK

The strength of DMRG lies in its treatment of identical sites that are energetically degenerate and spatially similar. However, this becomes a drawback when applied to quantum chemistry calculations for large systems, as entangled orbitals often span broad ranges in energy and space, with notably inhomogeneous interactions. In this study, we propose addressing strong intra-fragment and weak inter-fragment correlations separately using a multi-configurational block interaction product state (BIPS) framework. The strong correlation is captured in electronic states on fragments, considering entanglement between fragments and their environments. This method has been tested in various chemical systems and shows high accuracy and efficiency in addressing inhomogeneous effects in quantum chemistry.

We present a multi-frequency polarimetric study for the quasar 1604+159. The source was observed at the $L$ band with the American Very Long Baseline Array (VLBA) and the $L$, $X$, and $U$ bands with the Very Large Array (VLA). These observations provide different resolutions from mas to arcsec, enabling us to probe the morphology and magnetic field from tens of parsec to hundreds of kilo-parsec scale. We detect a symmetrical Fanaroff-Riley-Class-I-like structure. The source has several lobes and bulges, forming a cocoon shape. The polarization is normal to the edges of the structure with high fractional polarization up to $\sim 60\%$. Two hotspots are observed at the eastern and western sides of the source, located symmetrically relative to the core. The flux density ratio ($>1.5$) between the two hotspots suggests the Doppler beaming effect exists at a large scale. The polarized emission in the hotspots also shows a symmetrical structure with an oblique direction from the jet direction. In general, the jet propagates in a collimating structure with several bends. Polarization is also detected perpendicular to the local jet from $\sim$100 mas to $\sim$ 1 arcsec. The jet shows strong polarized intensity and high fractional polarization at the bending edges. We discuss the possible origins of the observed structure and magnetic field., Comment: 17 pages, accepted for publication in ApJ

Ultrawide bandgap semiconductor $\beta$-Ga2O3 holds extensive potential for applications in high-radiation environments. One of the primary challenges in its practical application is unveiling the mechanisms of surface irradiation damage under extreme conditions. In this study, we investigate the orientation-dependent mechanisms of radiation damage on four experimentally relevant $\beta$-Ga2O3 surface facets, namely, (100), (010), (001), and (-201), at various temperatures. We employ a multiscale atomic simulation approach, combining machine-learning-driven molecular dynamics (ML-MD) simulations and density functional theory (DFT) calculations. The results reveal that Ga vacancies and O interstitials are the predominant defects across all four surfaces, with the formation of many antisite defects Ga_O and few O_Ga observed. Among the two Ga sites and three O sites, the vacancy found in the O2 site is dominant, while the interstitials at the Ga1 and O1 sites are more significant. Interestingly, the (010) surface exhibits the lowest defect density, owing to its more profound channeling effect leading to a broader spread of defects. The influence of temperature on surface irradiation damage of $\beta$-Ga2O3 should be evaluated based on the unique crystal surface characteristics. Moreover, the formation energy and defect concentration calculated by DFT corroborate the results of the MD simulations. Comprehending surface radiation damage at the atomic level is crucial for assessing the radiation tolerance and predicting the performance changes of $\beta$-Ga2O3-based device in high-radiation environments.

Recent advancements in large language models (LLMs) have shown promise in generating psychotherapeutic dialogues, especially in Motivational Interviewing (MI). However, how to employ strategies, a set of motivational interviewing (MI) skills, to generate therapeutic-adherent conversations with explainability is underexplored. We propose an approach called strategy-aware dialogue generation with Chain-of-Strategy (CoS) planning, which first predicts MI strategies as reasoning and utilizes these strategies to guide the subsequent dialogue generation. It brings the potential for controllable and explainable generation in psychotherapy by aligning the generated MI dialogues with therapeutic strategies. Extensive experiments including automatic and human evaluations are conducted to validate the effectiveness of the MI strategy. Our findings demonstrate the potential of LLMs in producing strategically aligned dialogues and suggest directions for practical applications in psychotherapeutic settings.

Material responses to static and dynamic stimuli, represented as nonlinear curves, are design targets for engineering functionalities like structural support, impact protection, and acoustic and photonic bandgaps. Three-dimensional metamaterials offer significant tunability due to their internal structure, yet existing methods struggle to capture their complex behavior-to-structure relationships. We present GraphMetaMat, a graph-based framework capable of designing three-dimensional metamaterials with programmable responses and arbitrary manufacturing constraints. Integrating graph networks, physics biases, reinforcement learning, and tree search, GraphMetaMat can target stress-strain curves spanning four orders of magnitude and complex behaviors, as well as viscoelastic transmission responses with varying attenuation gaps. GraphMetaMat can create cushioning materials for protective equipment and vibration-damping panels for electric vehicles, outperforming commercial materials, and enabling the automatic design of materials with on-demand functionalities., Comment: 19 pages, 5 figures

Bounded by the size, weight, and power consumption (SWaP) of conventional superconducting single photon detectors (SSPD), applications of SSPDs were commonly confined in the laboratory. However, booming demands for high efficiency single photon detector incorporated with avionic platforms arise with the development of remote imaging and sensing or long-haul quantum communication without topographical constraints. We herein designed and manufactured the first drone based SSPD system with a SDE as high as 91.8%. This drone based SSPD system is established with high performance NbTiN SSPDs, self-developed miniature liquid helium dewar, and homemade integrated electric setups, which is able to be launched in complex topographical conditions. Such a drone based SSPD system may open the use of SSPDs for applications that demand high-SDE in complex environments.

Determining the John ellipsoid - the largest volume ellipsoid contained within a convex polytope - is a fundamental problem with applications in machine learning, optimization, and data analytics. Recent work has developed fast algorithms for approximating the John ellipsoid using sketching and leverage score sampling techniques. However, these algorithms do not provide privacy guarantees for sensitive input data. In this paper, we present the first differentially private algorithm for fast John ellipsoid computation. Our method integrates noise perturbation with sketching and leverage score sampling to achieve both efficiency and privacy. We prove that (1) our algorithm provides $(\epsilon,\delta)$-differential privacy, and the privacy guarantee holds for neighboring datasets that are $\epsilon_0$-close, allowing flexibility in the privacy definition; (2) our algorithm still converges to a $(1+\xi)$-approximation of the optimal John ellipsoid in $O(\xi^{-2}(\log(n/\delta_0) + (L\epsilon_0)^{-2}))$ iterations where $n$ is the number of data point, $L$ is the Lipschitz constant, $\delta_0$ is the failure probability, and $\epsilon_0$ is the closeness of neighboring input datasets. Our theoretical analysis demonstrates the algorithm's convergence and privacy properties, providing a robust approach for balancing utility and privacy in John ellipsoid computation. This is the first differentially private algorithm for fast John ellipsoid computation, opening avenues for future research in privacy-preserving optimization techniques.

Medical Image Foundation Models have proven to be powerful tools for mask prediction across various datasets. However, accurately assessing the uncertainty of their predictions remains a significant challenge. To address this, we propose a new model, U-MedSAM, which integrates the MedSAM model with an uncertainty-aware loss function and the Sharpness-Aware Minimization (SharpMin) optimizer. The uncertainty-aware loss function automatically combines region-based, distribution-based, and pixel-based loss designs to enhance segmentation accuracy and robustness. SharpMin improves generalization by finding flat minima in the loss landscape, thereby reducing overfitting. Our method was evaluated in the CVPR24 MedSAM on Laptop challenge, where U-MedSAM demonstrated promising performance.

Early assessment of tumor therapeutic response is an important topic in precision medicine to optimize personalized treatment regimens and reduce unnecessary toxicity, cost, and delay. Although diffusion MRI (dMRI) has shown potential to address this need, its predictive accuracy is limited, likely due to its unspecific sensitivity to overall pathological changes. In this work, we propose a new quantitative dMRI-based method dubbed EXCHANGE (MRI of water Exchange, Confined and Hindered diffusion under Arbitrary Gradient waveform Encodings) for simultaneous mapping of cell size, cell density, and transcytolemmal water exchange. Such rich microstructural information comprehensively evaluates tumor pathologies at the cellular level. Validations using numerical simulations and in vitro cell experiments confirmed that the EXCHANGE method can accurately estimate mean cell size, density, and water exchange rate constants. The results from in vivo animal experiments show the potential of EXCHANGE for monitoring tumor treatment response. Finally, the EXCHANGE method was implemented in breast cancer patients with neoadjuvant chemotherapy, demonstrating its feasibility in assessing tumor therapeutic response in clinics. In summary, a new, quantitative dMRI-based EXCHANGE method was proposed to comprehensively characterize tumor microstructural properties at the cellular level, suggesting a unique means to monitor tumor treatment response in clinical practice., Comment: 40 pages, 6 figures

Magic-angle twisted bilayer graphene (MATBG) hosts a multitude of strongly correlated states at partial fillings of its flat bands. In a magnetic field, these flat bands further evolve into a unique Hofstadter spectrum renormalized by strong Coulomb interactions. Here, we study the interacting Hofstadter states spontaneously formed within the topological magnetic subbands of an ultraclean MATBG device, notably including symmetry-broken Chern insulator (SBCI) states and fractional quantum Hall (FQH) states. The observed SBCI states form a cascade with their Chern numbers mimicking the main sequence correlated Chern insulators. The FQH states in MATBG form in Jain sequence; however, they disappear at high magnetic field, distinct from conventional FQH states which strengthen with increasing magnetic field. We reveal a unique magnetic field-driven phase transition from composite fermion phases to a dissipative Fermi liquid. Our theoretical analysis of the magnetic subbands hosting FQH states predicts non uniform quantum geometric properties far from the lowest Landau level. This points towards a more natural interpretation of these FQH states as in-field fractional Chern insulators of the magnetic subbands.

We introduce a new quantum layout-aware approach to realize cost-effective $n$-bit gates using the Bloch sphere, for $2 \le n \le 5$ qubits. These $n$-bit gates are entirely constructed from the Clifford+T gates, in the approach of selecting sequences of rotations visualized on the Bloch sphere. This Bloch sphere approach ensures to match the quantum layout for synthesizing (transpiling) these $n$-bit gates into an IBM quantum computer. Various standard $n$-bit gates (Toffoli, Fredkin, etc.) and their operational equivalent of our proposed $n$-bit gates are examined and evaluated, in the context of the final quantum costs, as the final counts of generated IBM native gates. In this paper, we demonstrate that all our $n$-bit gates always have lower quantum costs than those of standard $n$-bit gates after transpilation. Hence, our Bloch sphere approach can be used to build a quantum library of various cost-effective $n$-bit gates for different layouts of IBM quantum computers., Comment: 27 pages, 14 figures, and 8 tables

In this paper, we consider the following Choquard type equation \begin{equation} \left\{\begin{aligned} &-\Delta u+\lambda u=\gamma(\Phi_N(|x|)\ast|u|^p)u \ \ \mbox{in $\mathbb{R}^N$}, \\ &\lim\limits_{|x|\to\infty}u(x)=0,\\ \end{aligned}\right. \end{equation} where $N\geq2,\lambda>0,\gamma>0, p\in[1,2]$ and $\Phi_N(|x|)$ denotes the fundamental solution of the Laplacian $-\Delta$ on $\mathbb{R}^N$. This equation does not have a variational frame when $p\neq 2.$ Instead of variational methods, we prove the existence and uniqueness of positive radial solutions of the above equation via the shooting method by establishing some new differential inequalities. The proofs are based on an analysis of the corresponding system of second-order differential equations, and our results extend the existing ones in the literature from $p=2$ to $p\in[1,2]$., Comment: 18pages

Creating a 360{\deg} parametric model of a human head is a very challenging task. While recent advancements have demonstrated the efficacy of leveraging synthetic data for building such parametric head models, their performance remains inadequate in crucial areas such as expression-driven animation, hairstyle editing, and text-based modifications. In this paper, we build a dataset of artist-designed high-fidelity human heads and propose to create a novel parametric 360{\deg} renderable parametric head model from it. Our scheme decouples the facial motion/shape and facial appearance, which are represented by a classic parametric 3D mesh model and an attached neural texture, respectively. We further propose a training method for decompositing hairstyle and facial appearance, allowing free-swapping of the hairstyle. A novel inversion fitting method is presented based on single image input with high generalization and fidelity. To the best of our knowledge, our model is the first parametric 3D full-head that achieves 360{\deg} free-view synthesis, image-based fitting, appearance editing, and animation within a single model. Experiments show that facial motions and appearances are well disentangled in the parametric space, leading to SOTA performance in rendering and animating quality. The code and SynHead100 dataset are released at https://nju-3dv.github.io/projects/Head360., Comment: ECCV 2024

We introduce Berkeley Humanoid, a reliable and low-cost mid-scale humanoid research platform for learning-based control. Our lightweight, in-house-built robot is designed specifically for learning algorithms with low simulation complexity, anthropomorphic motion, and high reliability against falls. The robot's narrow sim-to-real gap enables agile and robust locomotion across various terrains in outdoor environments, achieved with a simple reinforcement learning controller using light domain randomization. Furthermore, we demonstrate the robot traversing for hundreds of meters, walking on a steep unpaved trail, and hopping with single and double legs as a testimony to its high performance in dynamical walking. Capable of omnidirectional locomotion and withstanding large perturbations with a compact setup, our system aims for scalable, sim-to-real deployment of learning-based humanoid systems. Please check http://berkeley-humanoid.com for more details., Comment: 12 pages, 9 figures

The demand for social robots in fields like healthcare, education, and entertainment increases due to their emotional adaptation features. These robots leverage multimodal communication, incorporating speech, facial expressions, and gestures to enhance user engagement and emotional support. The understanding of design paradigms of social robots is obstructed by the complexity of the system and the necessity to tune it to a specific task. This article provides a structured review of social robot design paradigms, categorizing them into cognitive architectures, role design models, linguistic models, communication flow, activity system models, and integrated design models. By breaking down the articles on social robot design and application based on these paradigms, we highlight the strengths and areas for improvement in current approaches. We further propose our original integrated design model that combines the most important aspects of the design of social robots. Our approach shows the importance of integrating operational, communicational, and emotional dimensions to create more adaptive and empathetic interactions between robots and humans.

Multiple Instance Learning (MIL) has garnered widespread attention in the field of Whole Slide Image (WSI) classification as it replaces pixel-level manual annotation with diagnostic reports as labels, significantly reducing labor costs. Recent research has shown that bag-level MIL methods often yield better results because they can consider all patches of the WSI as a whole. However, a drawback of such methods is the incorporation of more redundant patches, leading to interference. To extract patches with high diagnostic value while excluding interfering patches to address this issue, we developed an attention-based feature distillation multi-instance learning (AFD-MIL) approach. This approach proposed the exclusion of redundant patches as a preprocessing operation in weakly supervised learning, directly mitigating interference from extensive noise. It also pioneers the use of attention mechanisms to distill features with high diagnostic value, as opposed to the traditional practice of indiscriminately and forcibly integrating all patches. Additionally, we introduced global loss optimization to finely control the feature distillation module. AFD-MIL is orthogonal to many existing MIL methods, leading to consistent performance improvements. This approach has surpassed the current state-of-the-art method, achieving 91.47% ACC (accuracy) and 94.29% AUC (area under the curve) on the Camelyon16 (Camelyon Challenge 2016, breast cancer), while 93.33% ACC and 98.17% AUC on the TCGA-NSCLC (The Cancer Genome Atlas Program: non-small cell lung cancer). Different feature distillation methods were used for the two datasets, tailored to the specific diseases, thereby improving performance and interpretability.

The spin-orbit assisted Mott insulator $\alpha$-RuCl3 is proximate to the coveted quantum spin liquid (QSL) predicted by the Kitaev model. In the search for the pure Kitaev QSL, reducing the dimensionality of this frustrated magnet by exfoliation has been proposed as a way to enhance magnetic fluctuations and Kitaev interactions. Here, we perform angle-dependent tunneling magnetoresistance (TMR) measurements on ultrathin $\alpha$-RuCl3 crystals with various layer numbers to probe their magnetic, electronic and crystal structure. We observe a giant change in resistance - as large as ~2500% - when the magnetic field rotates either within or out of the $\alpha$-RuCl3 plane, a manifestation of the strongly anisotropic spin interactions in this material. In combination with scanning transmission electron microscopy, this tunneling anisotropic magnetoresistance (TAMR) reveals that few-layer $\alpha$-RuCl3 crystals remain in the high-temperature monoclinic phase at low temperature. It also shows the presence of a zigzag antiferromagnetic order below the critical temperature TN ~ 14 K, which is twice the one typically observed in bulk samples with rhombohedral stacking. Our work offers valuable insights into the relation between the stacking order and magnetic properties of this material, which helps lay the groundwork for creating and electrically probing exotic magnetic phases like QSLs via van der Waals engineering.

Integrating deep neural networks with the Hawkes process has significantly improved predictive capabilities in finance, health informatics, and information technology. Nevertheless, these models often face challenges in real-world settings, particularly due to substantial label noise. This issue is of significant concern in the medical field, where label noise can arise from delayed updates in electronic medical records or misdiagnoses, leading to increased prediction risks. Our research indicates that deep Hawkes process models exhibit reduced robustness when dealing with label noise, particularly when it affects both event types and timing. To address these challenges, we first investigate the influence of label noise in approximated intensity functions and present a novel framework, the Robust Deep Hawkes Process (RDHP), to overcome the impact of label noise on the intensity function of Hawkes models, considering both the events and their occurrences. We tested RDHP using multiple open-source benchmarks with synthetic noise and conducted a case study on obstructive sleep apnea-hypopnea syndrome (OSAHS) in a real-world setting with inherent label noise. The results demonstrate that RDHP can effectively perform classification and regression tasks, even in the presence of noise related to events and their timing. To the best of our knowledge, this is the first study to successfully address both event and time label noise in deep Hawkes process models, offering a promising solution for medical applications, specifically in diagnosing OSAHS., Comment: ECAI2024

In the field of deep learning-driven Whole Slide Image (WSI) classification, Multiple Instance Learning (MIL) has gained significant attention due to its ability to be trained using only slide-level diagnostic labels. Previous MIL researches have primarily focused on enhancing feature aggregators for globally analyzing WSIs, but overlook a causal relationship in diagnosis: model's prediction should ideally stem solely from regions of the image that contain diagnostic evidence (such as tumor cells), which usually occupy relatively small areas. To address this limitation and establish the truly causal relationship between model predictions and diagnostic evidence regions, we propose Causal Inference Multiple Instance Learning (CI-MIL). CI-MIL integrates feature distillation with a novel patch decorrelation mechanism, employing a two-stage causal inference approach to distill and process patches with high diagnostic value. Initially, CI-MIL leverages feature distillation to identify patches likely containing tumor cells and extracts their corresponding feature representations. These features are then mapped to random Fourier feature space, where a learnable weighting scheme is employed to minimize inter-feature correlations, effectively reducing redundancy from homogenous patches and mitigating data bias. These processes strengthen the causal relationship between model predictions and diagnostically relevant regions, making the prediction more direct and reliable. Experimental results demonstrate that CI-MIL outperforms state-of-the-art methods. Additionally, CI-MIL exhibits superior interpretability, as its selected regions demonstrate high consistency with ground truth annotations, promising more reliable diagnostic assistance for pathologists.

Large language models (LLMs) exhibit remarkable in-context learning (ICL) capabilities. However, the underlying working mechanism of ICL remains poorly understood. Recent research presents two conflicting views on ICL: One attributes it to LLMs' inherent ability of task recognition, deeming label correctness and shot numbers of demonstrations as not crucial; the other emphasizes the impact of similar examples in the demonstrations, stressing the need for label correctness and more shots. In this work, we provide a Two-Dimensional Coordinate System that unifies both views into a systematic framework. The framework explains the behavior of ICL through two orthogonal variables: whether LLMs can recognize the task and whether similar examples are presented in the demonstrations. We propose the peak inverse rank metric to detect the task recognition ability of LLMs and study LLMs' reactions to different definitions of similarity. Based on these, we conduct extensive experiments to elucidate how ICL functions across each quadrant on multiple representative classification tasks. Finally, we extend our analyses to generation tasks, showing that our coordinate system can also be used to interpret ICL for generation tasks effectively.

We derive the short-maturity asymptotics for European and VIX option prices in local-stochastic volatility models where the volatility follows a continuous-path Markov process. Both out-of-the-money (OTM) and at-the-money (ATM) asymptotics are considered. Using large deviations theory methods, the asymptotics for the OTM options are expressed as a two-dimensional variational problem, which is reduced to an extremal problem for a function of two real variables. This extremal problem is solved explicitly in an expansion in log-moneyness. We derive series expansions for the implied volatility for European and VIX options which should be useful for model calibration. We give explicit results for two classes of local-stochastic volatility models relevant in practice, with Heston-type and SABR-type stochastic volatility. The leading-order asymptotics for at-the-money options are computed in closed-form. The asymptotic results reproduce known results in the literature for the Heston and SABR models and for the uncorrelated local-stochastic volatility model. The asymptotic results are tested against numerical simulations for a local-stochastic volatility model with bounded local volatility., Comment: 58 pages, 2 figures, 1 table

The rapid development and widespread adoption of Generative Artificial Intelligence-based (GAI) applications have greatly enriched our daily lives, benefiting people by enhancing creativity, personalizing experiences, improving accessibility, and fostering innovation and efficiency across various domains. However, along with the development of GAI applications, concerns have been raised about transparency in their privacy practices. Traditional privacy policies often fail to effectively communicate essential privacy information due to their complexity and length, and open-source community developers often neglect privacy practices even more. Only 12.2% of examined open-source GAI apps provide a privacy policy. To address this, we propose a regulation-driven GAI Privacy Label and introduce Repo2Label, a novel framework for automatically generating these labels based on code repositories. Our user study indicates a common endorsement of the proposed GAI privacy label format. Additionally, Repo2Label achieves a precision of 0.81, recall of 0.88, and F1-score of 0.84 based on the benchmark dataset, significantly outperforming the developer self-declared privacy notices. We also discuss the common regulatory (in)compliance of open-source GAI apps, comparison with other privacy notices, and broader impacts to different stakeholders. Our findings suggest that Repo2Label could serve as a significant tool for bolstering the privacy transparency of GAI apps and make them more practical and responsible.

The natural van der Waals superlattice MnBi2Te4-(Bi2Te3)m provides an optimal platform to combine topology and magnetism in one system with minimal structural disorder. Here, we show that this system can harbor both ferromagnetic (FM) and antiferromagnetic (AFM) orders and that these magnetic orders can be controlled in two different ways by either varying the Mn-Mn distance while keeping the Bi2Te3/MnBi2Te4 ratio constant or vice versa. We achieve this by creating atomically engineered sandwich structures composed of Bi2Te3 and MnBi2Te4 layers. We show that the AFM order is exclusively determined by the Mn-Mn distance whereas the FM order depends only on the overall Bi2Te3/MnBi2Te4 ratio regardless of the distance between the MnBi2Te4 layers. Our results shed light on the origins of the AFM and FM orders and provide insights into how to manipulate magnetic orders not only for the MnBi2Te4-Bi2Te3 system but also for other magneto-topological materials., Comment: 25 pages, 5 figures, accepted to Nano Letters

Monolayer MoTe2 exhibits a variety of derivative structural phases and associated novel electronic properties that enable a wealth of potential applications in future electronic and optoelectronic devices. However, a comprehensive study focusing on the complexities of the controllable phase evolution in this atomically thin film has yet to be performed. This work aims to address this issue by systematically investigating molecular beam epitaxial growth of monolayer Mo-Te compounds on bilayer graphene substrates. By utilizing scanning tunnelling microscopy, we explored a series of thermally driven structural phase evolutions including distinct T'-MoTe2, H-MoTe2, Mo6Te6 nanowires, and multistoichiometric MoTe2-x. Furthermore, we carefully investigated the critical effects of the growth parameters-annealing temperature and time and tellurium concentration-on the controllable and reversible phase transformation within monolayer MoTe2-x. The findings have significant implications for understanding the thin film synthesis and phase transformation engineering inherent to two-dimensional crystals, which can foster further development of high-performance devices., Comment: 17 pages, 5 figures

Using questions in written text is an effective strategy to enhance readability. However, what makes an active reading question good, what the linguistic role of these questions is, and what is their impact on human reading remains understudied. We introduce GuidingQ, a dataset of 10K in-text questions from textbooks and scientific articles. By analyzing the dataset, we present a comprehensive understanding of the use, distribution, and linguistic characteristics of these questions. Then, we explore various approaches to generate such questions using language models. Our results highlight the importance of capturing inter-question relationships and the challenge of question position identification in generating these questions. Finally, we conduct a human study to understand the implication of such questions on reading comprehension. We find that the generated questions are of high quality and are almost as effective as human-written questions in terms of improving readers' memorization and comprehension., Comment: ACL 2024

Large language models (LLMs) have shown exceptional performance as general-purpose assistants, excelling across a variety of reasoning tasks. This achievement represents a significant step toward achieving artificial general intelligence (AGI). Despite these advancements, the effectiveness of LLMs often hinges on the specific prompting strategies employed, and there remains a lack of a robust framework to facilitate learning and generalization across diverse reasoning tasks. To address these challenges, we introduce a novel learning framework, THOUGHT-LIKE-PRO In this framework, we utilize imitation learning to imitate the Chain-of-Thought (CoT) process which is verified and translated from reasoning trajectories generated by a symbolic Prolog logic engine. This framework proceeds in a self-driven manner, that enables LLMs to formulate rules and statements from given instructions and leverage the symbolic Prolog engine to derive results. Subsequently, LLMs convert Prolog-derived successive reasoning trajectories into natural language CoT for imitation learning. Our empirical findings indicate that our proposed approach substantially enhances the reasoning abilities of LLMs and demonstrates robust generalization across out-of-distribution reasoning tasks.

The rapid development of spatial transcriptomics (ST) technologies is revolutionizing our understanding of the spatial organization of biological tissues. Current ST methods, categorized into next-generation sequencing-based (seq-based) and fluorescence in situ hybridization-based (image-based) methods, offer innovative insights into the functional dynamics of biological tissues. However, these methods are limited by their cellular resolution and the quantity of genes they can detect. To address these limitations, we propose SpaDiT, a deep learning method that utilizes a diffusion generative model to integrate scRNA-seq and ST data for the prediction of undetected genes. By employing a Transformer-based diffusion model, SpaDiT not only accurately predicts unknown genes but also effectively generates the spatial structure of ST genes. We have demonstrated the effectiveness of SpaDiT through extensive experiments on both seq-based and image-based ST data. SpaDiT significantly contributes to ST gene prediction methods with its innovative approach. Compared to eight leading baseline methods, SpaDiT achieved state-of-the-art performance across multiple metrics, highlighting its substantial bioinformatics contribution.

Recently, text-to-3D generation has attracted significant attention, resulting in notable performance enhancements. Previous methods utilize end-to-end 3D generation models to initialize 3D Gaussians, multi-view diffusion models to enforce multi-view consistency, and text-to-image diffusion models to refine details with score distillation algorithms. However, these methods exhibit two limitations. Firstly, they encounter conflicts in generation directions since different models aim to produce diverse 3D assets. Secondly, the issue of over-saturation in score distillation has not been thoroughly investigated and solved. To address these limitations, we propose PlacidDreamer, a text-to-3D framework that harmonizes initialization, multi-view generation, and text-conditioned generation with a single multi-view diffusion model, while simultaneously employing a novel score distillation algorithm to achieve balanced saturation. To unify the generation direction, we introduce the Latent-Plane module, a training-friendly plug-in extension that enables multi-view diffusion models to provide fast geometry reconstruction for initialization and enhanced multi-view images to personalize the text-to-image diffusion model. To address the over-saturation problem, we propose to view score distillation as a multi-objective optimization problem and introduce the Balanced Score Distillation algorithm, which offers a Pareto Optimal solution that achieves both rich details and balanced saturation. Extensive experiments validate the outstanding capabilities of our PlacidDreamer. The code is available at \url{https://github.com/HansenHuang0823/PlacidDreamer}., Comment: Accepted by ACM Multimedia 2024

In this paper, we investigate the use of diffusion models which are pre-trained on large-scale image-caption pairs for open-vocabulary 3D semantic understanding. We propose a novel method, namely Diff2Scene, which leverages frozen representations from text-image generative models, along with salient-aware and geometric-aware masks, for open-vocabulary 3D semantic segmentation and visual grounding tasks. Diff2Scene gets rid of any labeled 3D data and effectively identifies objects, appearances, materials, locations and their compositions in 3D scenes. We show that it outperforms competitive baselines and achieves significant improvements over state-of-the-art methods. In particular, Diff2Scene improves the state-of-the-art method on ScanNet200 by 12%., Comment: ECCV 2024

Effective collaboration in multi-agent systems requires communicating goals and intentions between agents. Current agent frameworks often suffer from dependencies on single-agent execution and lack robust inter-module communication, frequently leading to suboptimal multi-agent reinforcement learning (MARL) policies and inadequate task coordination. To address these challenges, we present a framework for training large language models (LLMs) as collaborative agents to enable coordinated behaviors in cooperative MARL. Each agent maintains a private intention consisting of its current goal and associated sub-tasks. Agents broadcast their intentions periodically, allowing other agents to infer coordination tasks. A propagation network transforms broadcast intentions into teammate-specific communication messages, sharing relevant goals with designated teammates. The architecture of our framework is structured into planning, grounding, and execution modules. During execution, multiple agents interact in a downstream environment and communicate intentions to enable coordinated behaviors. The grounding module dynamically adapts comprehension strategies based on emerging coordination patterns, while feedback from execution agents influnces the planning module, enabling the dynamic re-planning of sub-tasks. Results in collaborative environment simulation demonstrate intention propagation reduces miscoordination errors by aligning sub-task dependencies between agents. Agents learn when to communicate intentions and which teammates require task details, resulting in emergent coordinated behaviors. This demonstrates the efficacy of intention sharing for cooperative multi-agent RL based on LLMs.

Structured data, rich in logical and relational information, has the potential to enhance the reasoning abilities of large language models (LLMs). Still, its integration poses a challenge due to the risk of overwhelming LLMs with excessive tokens and irrelevant context information. To address this, we propose Struct-X, a novel framework that operates through five key phases: ``read-model-fill-reflect-reason'' efficiently enabling LLMs to utilize structured data. It begins by encoding structured data into a topological space using graph embeddings, followed by filling in missing entity information with knowledge retrieval modules, and filtering out irrelevant tokens via a self-supervised module. The final phase involves constructing a topological network with selected tokens to further reduce the total token length for more effective LLM inference. Additionally, Struct-X includes an Auxiliary Module trained to generate prompts, aiding LLMs in analyzing structured data. Extensive experiments on benchmarks, including the knowledge graph question-answer task and the long document reading comprehension task, show that Struct-X notably improves LLM reasoning, demonstrating the effectiveness of structured data augmentation in improving LLM inference with complex input context.

Latest advances have achieved realistic virtual try-on (VTON) through localized garment inpainting using latent diffusion models, significantly enhancing consumers' online shopping experience. However, existing VTON technologies neglect the need for merchants to showcase garments comprehensively, including flexible control over garments, optional faces, poses, and scenes. To address this issue, we define a virtual dressing (VD) task focused on generating freely editable human images with fixed garments and optional conditions. Meanwhile, we design a comprehensive affinity metric index (CAMI) to evaluate the consistency between generated images and reference garments. Then, we propose IMAGDressing-v1, which incorporates a garment UNet that captures semantic features from CLIP and texture features from VAE. We present a hybrid attention module, including a frozen self-attention and a trainable cross-attention, to integrate garment features from the garment UNet into a frozen denoising UNet, ensuring users can control different scenes through text. IMAGDressing-v1 can be combined with other extension plugins, such as ControlNet and IP-Adapter, to enhance the diversity and controllability of generated images. Furthermore, to address the lack of data, we release the interactive garment pairing (IGPair) dataset, containing over 300,000 pairs of clothing and dressed images, and establish a standard pipeline for data assembly. Extensive experiments demonstrate that our IMAGDressing-v1 achieves state-of-the-art human image synthesis performance under various controlled conditions. The code and model will be available at https://github.com/muzishen/IMAGDressing.

The task of labeling multiple organs for segmentation is a complex and time-consuming process, resulting in a scarcity of comprehensively labeled multi-organ datasets while the emergence of numerous partially labeled datasets. Current methods are inadequate in effectively utilizing the supervised information available from these datasets, thereby impeding the progress in improving the segmentation accuracy. This paper proposes a two-stage multi-organ segmentation method based on mutual learning, aiming to improve multi-organ segmentation performance by complementing information among partially labeled datasets. In the first stage, each partial-organ segmentation model utilizes the non-overlapping organ labels from different datasets and the distinct organ features extracted by different models, introducing additional mutual difference learning to generate higher quality pseudo labels for unlabeled organs. In the second stage, each full-organ segmentation model is supervised by fully labeled datasets with pseudo labels and leverages true labels from other datasets, while dynamically sharing accurate features across different models, introducing additional mutual similarity learning to enhance multi-organ segmentation performance. Extensive experiments were conducted on nine datasets that included the head and neck, chest, abdomen, and pelvis. The results indicate that our method has achieved SOTA performance in segmentation tasks that rely on partial labels, and the ablation studies have thoroughly confirmed the efficacy of the mutual learning mechanism., Comment: 10 pages, 9 figures

The X$_2$MH$_6$ family, consisting of an electropositive cation X and a main group metal M octahedrally coordinated by hydrogen, has been predicted to hold promise for high-temperature conventional superconductivity. Herein, we analyze the electronic structure of two members of this family, Mg$_2$IrH$_6$ and Ca$_2$IrH$_6$, showing why the former may possess superconducting properties rivaling those of the cuprates, whereas the latter does not. Within Mg$_2$IrH$_6$ the vibrations of the IrH$_6^{4-}$ anions are key for the superconducting mechanism, and they induce coupling in the $e_g^*$ set, which are antibonding between the H 1$s$ and the Ir $d_{x^2-y^2}$ or $d_{z^2}$ orbitals. Because calcium possesses low-lying d-orbitals, $e_g^* \rightarrow$ Ca $d$ back-donation is preferred, quenching the superconductivity. Our analysis explains why high critical temperatures were only predicted for second or third row X metal atoms, and may hold implications for superconductivity in other systems where the antibonding anionic states are filled.

Knowledge distillation plays a key role in compressing the Large Language Models (LLMs), which boosts a small-size student model under large teacher models' guidance. However, existing LLM distillation methods overly rely on student-generated outputs, which may introduce generation errors and misguide the distillation process. Moreover, the distillation loss functions introduced in previous art struggle to align the most informative part due to the complex distribution of LLMs' outputs. To address these problems, we propose a multi-granularity semantic revision method for LLM distillation. At the sequence level, we propose a sequence correction and re-generation (SCRG) strategy. SCRG first calculates the semantic cognitive difference between the teacher and student to detect the error token, then corrects it with the teacher-generated one, and re-generates the sequence to reduce generation errors and enhance generation diversity. At the token level, we design a distribution adaptive clipping Kullback-Leibler (DAC-KL) loss as the distillation objective function. DAC-KL loss exploits a learnable sub-network to adaptively extract semantically dense areas from the teacher's output, avoiding the interference of redundant information in the distillation process. Finally, at the span level, we leverage the span priors of a sequence to compute the probability correlations within spans, and constrain the teacher and student's probability correlations to be consistent, further enhancing the transfer of semantic information. Extensive experiments across different model families with parameters ranging from 0.1B to 13B demonstrate the superiority of our method compared to existing methods.

Storyboarding is an established method for designing user experiences. Generative AI can support this process by helping designers quickly create visual narratives. However, existing tools only focus on accurate text-to-image generation. Currently, it is not clear how to effectively support the entire creative process of storyboarding and how to develop AI-powered tools to support designers' individual workflows. In this work, we iteratively developed and implemented StoryDiffusion, a system that integrates text-to-text and text-to-image models, to support the generation of narratives and images in a single pipeline. With a user study, we observed 12 UX designers using the system for both concept ideation and illustration tasks. Our findings identified AI-directed vs. user-directed creative strategies in both tasks and revealed the importance of supporting the interchange between narrative iteration and image generation. We also found effects of the design tasks on their strategies and preferences, providing insights for future development.

This work presents HiLMa-Res, a hierarchical framework leveraging reinforcement learning to tackle manipulation tasks while performing continuous locomotion using quadrupedal robots. Unlike most previous efforts that focus on solving a specific task, HiLMa-Res is designed to be general for various loco-manipulation tasks that require quadrupedal robots to maintain sustained mobility. The novel design of this framework tackles the challenges of integrating continuous locomotion control and manipulation using legs. It develops an operational space locomotion controller that can track arbitrary robot end-effector (toe) trajectories while walking at different velocities. This controller is designed to be general to different downstream tasks, and therefore, can be utilized in high-level manipulation planning policy to address specific tasks. To demonstrate the versatility of this framework, we utilize HiLMa-Res to tackle several challenging loco-manipulation tasks using a quadrupedal robot in the real world. These tasks span from leveraging state-based policy to vision-based policy, from training purely from the simulation data to learning from real-world data. In these tasks, HiLMa-Res shows better performance than other methods., Comment: IROS 2024

CLIP has achieved impressive zero-shot performance after pre-training on a large-scale dataset consisting of paired image-text data. Previous works have utilized CLIP by incorporating manually designed visual prompts like colored circles and blur masks into the images to guide the model's attention, showing enhanced zero-shot performance in downstream tasks. Although these methods have achieved promising results, they inevitably alter the original information of the images, which can lead to failure in specific tasks. We propose a train-free method Foveal-Attention CLIP (FALIP), which adjusts the CLIP's attention by inserting foveal attention masks into the multi-head self-attention module. We demonstrate FALIP effectively boosts CLIP zero-shot performance in tasks such as referring expressions comprehension, image classification, and 3D point cloud recognition. Experimental results further show that FALIP outperforms existing methods on most metrics and can augment current methods to enhance their performance., Comment: Accepted by ECCV 2024, code released