Search

We prove Weyl laws for Schr\"odinger operators with critically singular potentials on compact manifolds with boundary. We also improve the Weyl remainder estimates under the condition that the set of all periodic geodesic billiards has measure 0. These extend the classical results by Seeley, Ivrii and Melrose. The proof uses the Gaussian heat kernel bounds for short times and a perturbation argument involving the wave equation., Comment: 19 pages

Automated breast tumor segmentation on the basis of dynamic contrast-enhancement magnetic resonance imaging (DCE-MRI) has shown great promise in clinical practice, particularly for identifying the presence of breast disease. However, accurate segmentation of breast tumor is a challenging task, often necessitating the development of complex networks. To strike an optimal trade-off between computational costs and segmentation performance, we propose a hybrid network via the combination of convolution neural network (CNN) and transformer layers. Specifically, the hybrid network consists of a encoder-decoder architecture by stacking convolution and decovolution layers. Effective 3D transformer layers are then implemented after the encoder subnetworks, to capture global dependencies between the bottleneck features. To improve the efficiency of hybrid network, two parallel encoder subnetworks are designed for the decoder and the transformer layers, respectively. To further enhance the discriminative capability of hybrid network, a prototype learning guided prediction module is proposed, where the category-specified prototypical features are calculated through on-line clustering. All learned prototypical features are finally combined with the features from decoder for tumor mask prediction. The experimental results on private and public DCE-MRI datasets demonstrate that the proposed hybrid network achieves superior performance than the state-of-the-art (SOTA) methods, while maintaining balance between segmentation accuracy and computation cost. Moreover, we demonstrate that automatically generated tumor masks can be effectively applied to identify HER2-positive subtype from HER2-negative subtype with the similar accuracy to the analysis based on manual tumor segmentation. The source code is available at https://github.com/ZhouL-lab/PLHN.

We have studied the leptonic EDM and the LFV decays relating with the recent data of anomalous magnetic moment of muon, $(g-2)_{\mu}$, in the framework of the SMEFT with the Stringy Ansatz. We have adopted the successful $\Gamma_2$ modular invariant model by Meloni-Parriciatu as the flavor symmetry of leptons. Suppose the anomaly of $(g-2)_{\mu}$, $\Delta a_{\mu}$ to be evidence of New Physics (NP), we have related it with the anomalous magnetic moment of the electron $\Delta a_e$, the electron EDM $d_e$ and the $\mu\to e \gamma$ decay. We found that the NP contributions to $\Delta a_{e(\mu)}$ are proportional to the lepton masses squared likewise the naive scaling $\Delta a_\ell \propto m^2_\ell$. The experimental constraint of $|d_e|$ is much tight compared with the one from the branching ratio $\mathcal{B}(\mu \to e \gamma)$ in our framework. Supposing the phase of our model parameter $\delta_{\alpha}$ for the electron to be of order one, we have estimated the upper-bound of $\mathcal{B}(\mu \to e \gamma)$, which is at most $10^{-21}-10^{-20}$. If the model parameters $\delta_{\alpha}$, $\delta_{\beta}$ (for muon) and $\delta_{\gamma}$ (for tauon) are real, leptonic EDMs vanish since the CP phase of the modular form due to modulus $\tau$ does not contribute to the EDM. However, we can obtain $\mathcal{B}(\mu \to e \gamma)\simeq 10^{-13}$ with non-vanishing $d_e$, that is the case, only $\delta_{\alpha}$ is real while the others are complex. Then, the imaginary part of $\delta_{\beta}$ can lead to $d_e$ in the next-to-leading contribution. The predicted electron EDM is below $10^{-32}$e\,cm, while $\mathcal{B}(\mu \to e \gamma)$ is close to the experimental upper-bound. The predicted upper-bound of $|d_\mu|$ is rather large, around $10^{-23}$e\,cm and of $|d_\tau|$ is $ 10^{-22}$e\,cm. The branching ratios of $\tau\to e\gamma$ and $\tau\to \mu\gamma$ are also discussed., Comment: 37 pages, 12 figures. arXiv admin note: text overlap with arXiv:2310.16325

Burq-G\'erard-Tzvetkov and Hu established $L^p$ estimates for the restriction of Laplace-Beltrami eigenfunctions to submanifolds. We investigate the eigenfunctions of the Schr\"odinger operators with critically singular potentials, and estimate the $L^p$ norms and period integrals for their restriction to submanifolds. Recently, Blair-Sire-Sogge obtained global $L^p$ bounds for Schr\"odinger eigenfunctions by the resolvent method. Due to the Sobolev trace inequalities, the resolvent method cannot work for submanifolds of all dimensions. We get around this difficulty and establish spectral projection bounds by the wave kernel techniques and the bootstrap argument involving an induction on the dimensions of the submanifolds., Comment: 36 pages, 1 figure

Scientific workflows are used to analyze large amounts of data. These workflows comprise numerous tasks, many of which are executed repeatedly, running the same custom program on different inputs. Users specify resource allocations for each task, which must be sufficient for all inputs to prevent task failures. As a result, task memory allocations tend to be overly conservative, wasting precious cluster resources, limiting overall parallelism, and increasing workflow makespan. In this paper, we first benchmark a state-of-the-art method on four real-life workflows from the nf-core workflow repository. This analysis reveals that certain assumptions underlying current prediction methods, which typically were evaluated only on simulated workflows, cannot generally be confirmed for real workflows and executions. We then present Ponder, a new online task-sizing strategy that considers and chooses between different methods to cater to different memory demand patterns. We implemented Ponder for Nextflow and made the code publicly available. In an experimental evaluation that also considers the impact of memory predictions on scheduling, Ponder improves Memory Allocation Quality on average by 71.0% and makespan by 21.8% in comparison to a state-of-the-art method. Moreover, Ponder produces 93.8% fewer task failures., Comment: Accepted at eScience'24

We study a variation of Kac's question, "Can one hear the shape of a drum?" if we allow ourselves access to some additional information. In particular, we allow ourselves to ``hear" the local Weyl counting function at each point on the manifold and ask if this is enough to uniquely recover the Riemannian metric. This is physically equivalent to asking whether one can determine the shape of a drum if one is allowed to knock at any place on the drum. We show that the answer to this question is ``yes" provided the Laplace-Beltrami spectrum of the drum is simple. We also provide a counterexample illustrating why this hypothesis is necessary., Comment: 7 pages

Al$_x$Ga$_{1-x}$N materials have become increasingly important for electronics in radiation environments due to their robust properties. In this work, we aim to investigate the atomistic mechanisms of radiation-induced damage in AlGaN compounds, providing insights that bridge the gap between high-length-scale experimental data and detailed atomic-level processes. Through extensive molecular dynamics simulations, we reveal the compositional dependence of radiation-induced defect production in Al$_x$Ga$_{1-x}$N systems with $x$ ranging from 0 to 1. The damage accumulation characteristics observed in our simulations align notably well with available experimental data at temperatures up to room temperature. Our findings indicate that alloy composition significantly influences defect production and microstructural evolution, including the formation of dislocation loops and defect clusters. Specifically, with increasing Al content, defect production from individual recoil events is reduced; however, extended interstitial defects are more likely to form considering cumulative effects, leading to enhanced damage at high doses. Among the compositions studied, we find that 25\% Al content leads to the least overall radiation damage, suggesting an optimal alloying strategy for mitigating radiation effects. These findings underscore the interplay between defect formation, dynamic annealing, and cascade effects, offering insights for optimizing AlGaN materials for radiation resistance in practical applications., Comment: 31 pages including 8 pages of references, 8 figures

It is well-known that all Feynman integrals within a given family can be expressed as a finite linear combination of master integrals. The master integrals naturally group into sectors. Starting from two loops, there can exist sectors made up of more than one master integral. In this paper we show that such sectors may have additional symmetries. First of all, self-duality, which was first observed in Feynman integrals related to Calabi--Yau geometries, often carries over to non-Calabi--Yau Feynman integrals. Secondly, we show that in addition there can exist Galois symmetries relating integrals. In the simplest case of two master integrals within a sector, whose definition involves a square root $r$, we may choose a basis $(I_1,I_2)$ such that $I_2$ is obtained from $I_1$ by the substitution $r \rightarrow -r$. This pattern also persists in sectors, which a priori are not related to any square root with dependence on the kinematic variables. We show in several examples that in such cases a suitable redefinition of the integrals introduces constant square roots like $\sqrt{3}$. The new master integrals are then again related by a Galois symmetry, for example the substitution $\sqrt{3} \rightarrow -\sqrt{3}$. To handle the case where the argument of a square root would be a perfect square we introduce a limit Galois symmetry. Both self-duality and Galois symmetries constrain the differential equation., Comment: 41 pages, v2: version to be published

Modelling complex deformation for soft robotics provides a guideline to understand their behaviour, leading to safe interaction with the environment. However, building a surrogate model with high accuracy and fast inference speed can be challenging for soft robotics due to the nonlinearity from complex geometry, large deformation, material nonlinearity etc. The reality gap from surrogate models also prevents their further deployment in the soft robotics domain. In this study, we proposed a physics-informed Neural Networks (PINNs) named PINN-Ray to model complex deformation for a Fin Ray soft robotic gripper, which embeds the minimum potential energy principle from elastic mechanics and additional high-fidelity experimental data into the loss function of neural network for training. This method is significant in terms of its generalisation to complex geometry and robust to data scarcity as compared to other data-driven neural networks. Furthermore, it has been extensively evaluated to model the deformation of the Fin Ray finger under external actuation. PINN-Ray demonstrates improved accuracy as compared with Finite element modelling (FEM) after applying the data assimilation scheme to treat the sim-to-real gap. Additionally, we introduced our automated framework to design, fabricate soft robotic fingers, and characterise their deformation by visual tracking, which provides a guideline for the fast prototype of soft robotics.

We introduce the sequence classification problem CIViC Evidence to the field of medical NLP. CIViC Evidence denotes the multi-label classification problem of assigning labels of clinical evidence to abstracts of scientific papers which have examined various combinations of genomic variants, cancer types, and treatment approaches. We approach CIViC Evidence using different language models: We fine-tune pretrained checkpoints of BERT and RoBERTa on the CIViC Evidence dataset and challenge their performance with models of the same architecture which have been pretrained on domain-specific text. In this context, we find that BiomedBERT and BioLinkBERT can outperform BERT on CIViC Evidence (+0.8% and +0.9% absolute improvement in class-support weighted F1 score). All transformer-based models show a clear performance edge when compared to a logistic regression trained on bigram tf-idf scores (+1.5 - 2.7% improved F1 score). We compare the aforementioned BERT-like models to OpenAI's GPT-4 in a few-shot setting (on a small subset of our original test dataset), demonstrating that, without additional prompt-engineering or fine-tuning, GPT-4 performs worse on CIViC Evidence than our six fine-tuned models (66.1% weighted F1 score compared to 71.8% for the best fine-tuned model). However, performance gets reasonably close to the benchmark of a logistic regression model trained on bigram tf-idf scores (67.7% weighted F1 score).

This paper investigates the cross-lingual inconsistencies observed in Large Language Models (LLMs), such as ChatGPT, Llama, and Baichuan, which have shown exceptional performance in various Natural Language Processing (NLP) tasks. Despite their successes, these models often exhibit significant inconsistencies when processing the same concepts across different languages. This study focuses on three primary questions: the existence of cross-lingual inconsistencies in LLMs, the specific aspects in which these inconsistencies manifest, and the correlation between cross-lingual consistency and multilingual capabilities of LLMs.To address these questions, we propose an innovative evaluation method for Cross-lingual Semantic Consistency (xSC) using the LaBSE model. We further introduce metrics for Cross-lingual Accuracy Consistency (xAC) and Cross-lingual Timeliness Consistency (xTC) to comprehensively assess the models' performance regarding semantic, accuracy, and timeliness inconsistencies. By harmonizing these metrics, we provide a holistic measurement of LLMs' cross-lingual consistency. Our findings aim to enhance the understanding and improvement of multilingual capabilities and interpretability in LLMs, contributing to the development of more robust and reliable multilingual language models.

In this paper we investigate the scattering amplitudes of spin-2 Kaluza-Klein (KK) states in Randall-Sundrum models with brane-localized curvature terms. We show that the presence of brane-localized curvature interactions modifies the properties of (4D) scalar fluctuations of the metric, resulting in scattering amplitudes of the massive spin-2 KK states which grow as ${\cal O}(s^3)$ instead of ${\cal O}(s)$. We discuss the constraints on the size of the brane-localized curvature interactions based on the consistency of the Sturm-Liouville mode systems of the spin-2 and spin-0 metric fluctuations. We connect the properties of the scattering amplitudes to the diffeomorphism invariance of the compactified KK theory with brane-localized curvature interactions. We verify that the scattering amplitudes involving brane-localized external sources (matter) are diffeomorphism-invariant, but show that those for matter localized at an arbitrary point in the bulk are not. We demonstrate that, in Feynman gauge, the spin-0 Goldstone bosons corresponding to helicity-0 states of the massive spin-2 KK bosons behave as a tower of Galileons, and that it is their interactions that produce the high-energy behavior of the scattering amplitudes. We also outline the correspondence between our results and those in the Dvali-Gabadadze-Porrati (DGP) model. In an appendix we discuss the analogous issue in extra-dimensional gauge theory, and show that the presence of a brane-localized gauge kinetic-energy term does not change the high-energy behavior of corresponding KK vector boson scattering amplitudes., Comment: 36 pages, 2 figures. Minor changes, new reference added

Liver cancer has a high incidence rate, but primary healthcare settings often lack experienced doctors. Advances in large models and AI technologies offer potential assistance. This work aims to address limitations in liver cancer diagnosis models, such as poor understanding of medical images, insufficient consideration of liver blood vessels, and ensuring accurate medical information. We propose a specialized diagnostic assistant to improve the diagnostic capabilities of less experienced doctors. Our framework combines large and small models, using optimized small models for precise patient image perception. Specifically, a segmentation network iteratively removes ambiguous pixels for liver tumor segmentation, and a multi-scale, multi-level differential network segments liver vessels. Features from these segmentations and medical records form a patient's personalized knowledge base. For diagnosis, Chain of Thought (COT) technology designs prompts mimicking experienced doctors' thought patterns, and Retrieval-Augmented Generation (RAG) technology provides answers based on reliable domain knowledge and trusted cases. Our small model methods improve liver tumor and vessel segmentation performance, resulting in more accurate information extraction. The large model component scores over 1 point higher on a 10-point scale in evaluations by doctors compared to control methods. Our method enhances semantic perception of medical images, improves classification of ambiguous pixels, and optimizes small object perception. It considers blood vessel positions for specific treatments and improves response credibility and interpretability by mimicking experienced doctors' thought processes using reliable resources. This approach has been recognized by doctors and benefits liver cancer auxiliary diagnosis.

Existing LLMs exhibit remarkable performance on various NLP tasks, but still struggle with complex real-world tasks, even equipped with advanced strategies like CoT and ReAct. In this work, we propose the CoAct framework, which transfers the hierarchical planning and collaboration patterns in human society to LLM systems. Specifically, our CoAct framework involves two agents: (1) A global planning agent, to comprehend the problem scope, formulate macro-level plans and provide detailed sub-task descriptions to local execution agents, which serves as the initial rendition of a global plan. (2) A local execution agent, to operate within the multi-tier task execution structure, focusing on detailed execution and implementation of specific tasks within the global plan. Experimental results on the WebArena benchmark show that CoAct can re-arrange the process trajectory when facing failures, and achieves superior performance over baseline methods on long-horizon web tasks. Code is available at https://github.com/xmhou2002/CoAct., Comment: 9 pages, 4 figures

Gloss-free sign language translation (SLT) aims to develop well-performing SLT systems with no requirement for the costly gloss annotations, but currently still lags behind gloss-based approaches significantly. In this paper, we identify a representation density problem that could be a bottleneck in restricting the performance of gloss-free SLT. Specifically, the representation density problem describes that the visual representations of semantically distinct sign gestures tend to be closely packed together in feature space, which makes gloss-free methods struggle with distinguishing different sign gestures and suffer from a sharp performance drop. To address the representation density problem, we introduce a simple but effective contrastive learning strategy, namely SignCL, which encourages gloss-free models to learn more discriminative feature representation in a self-supervised manner. Our experiments demonstrate that the proposed SignCL can significantly reduce the representation density and improve performance across various translation frameworks. Specifically, SignCL achieves a significant improvement in BLEU score for the Sign Language Transformer and GFSLT-VLP on the CSL-Daily dataset by 39% and 46%, respectively, without any increase of model parameters. Compared to Sign2GPT, a state-of-the-art method based on large-scale pre-trained vision and language models, SignCL achieves better performance with only 35% of its parameters. Implementation and Checkpoints are available at https://github.com/JinhuiYE/SignCL., Comment: Representation Density and Performance Drop

Recently, a series of diffusion-aware distillation algorithms have emerged to alleviate the computational overhead associated with the multi-step inference process of Diffusion Models (DMs). Current distillation techniques often dichotomize into two distinct aspects: i) ODE Trajectory Preservation; and ii) ODE Trajectory Reformulation. However, these approaches suffer from severe performance degradation or domain shifts. To address these limitations, we propose Hyper-SD, a novel framework that synergistically amalgamates the advantages of ODE Trajectory Preservation and Reformulation, while maintaining near-lossless performance during step compression. Firstly, we introduce Trajectory Segmented Consistency Distillation to progressively perform consistent distillation within pre-defined time-step segments, which facilitates the preservation of the original ODE trajectory from a higher-order perspective. Secondly, we incorporate human feedback learning to boost the performance of the model in a low-step regime and mitigate the performance loss incurred by the distillation process. Thirdly, we integrate score distillation to further improve the low-step generation capability of the model and offer the first attempt to leverage a unified LoRA to support the inference process at all steps. Extensive experiments and user studies demonstrate that Hyper-SD achieves SOTA performance from 1 to 8 inference steps for both SDXL and SD1.5. For example, Hyper-SDXL surpasses SDXL-Lightning by +0.68 in CLIP Score and +0.51 in Aes Score in the 1-step inference., Comment: Project Page: https://hyper-sd.github.io/

In this paper, we introduce StreakNet-Arch, a novel signal processing architecture designed for Underwater Carrier LiDAR-Radar (UCLR) imaging systems, to address the limitations in scatter suppression and real-time imaging. StreakNet-Arch formulates the signal processing as a real-time, end-to-end binary classification task, enabling real-time image acquisition. To achieve this, we leverage Self-Attention networks and propose a novel Double Branch Cross Attention (DBC-Attention) mechanism that surpasses the performance of traditional methods. Furthermore, we present a method for embedding streak-tube camera images into attention networks, effectively acting as a learned bandpass filter. To facilitate further research, we contribute a publicly available streak-tube camera image dataset. The dataset contains 2,695,168 real-world underwater 3D point cloud data. These advancements significantly improve UCLR capabilities, enhancing its performance and applicability in underwater imaging tasks. The source code and dataset can be found at https://github.com/BestAnHongjun/StreakNet ., Comment: Reduce the number of pages to 13

It has long been known that the maximal cut of the equal-mass four-loop banana integral is a period of a family of Calabi-Yau threefolds that depends on the kinematic variable $z=m^2/p^2$. We show that it can also be interpreted as a period of a family of genus-two curves. We do this by introducing a general Calabi-Yau-to-curve correspondence, which in this case locally relates the original period of the family of Calabi-Yau threefolds to a period of a family of genus-two curves that varies holomorphically with the kinematic variable $z$. In addition to working out the concrete details of this correspondence for the equal-mass four-loop banana integral, we outline when we expect a correspondence of this type to hold.

In the Letter, Xu et al. reported that edge modes disappear in the expanded structure of Wu-Hu model characterized by $\mathbb{Z}_2$ topological index, while appear in the trivial shrunken structure, when the edge cuts through the hexagonal unit cell. They then concluded that these edge modes are defect modes lacking topological protection. Unfortunately, their approach is not justified, rendering the conclusion unsolid., Comment: Comment on arXiv:2303.12617; To appear in Phys. Rev. Lett

The choice of a graph learning (GL) model (i.e., a GL algorithm and its hyperparameter settings) has a significant impact on the performance of downstream tasks. However, selecting the right GL model becomes increasingly difficult and time consuming as more and more GL models are developed. Accordingly, it is of great significance and practical value to equip users of GL with the ability to perform a near-instantaneous selection of an effective GL model without manual intervention. Despite the recent attempts to tackle this important problem, there has been no comprehensive benchmark environment to evaluate the performance of GL model selection methods. To bridge this gap, we present GLEMOS in this work, a comprehensive benchmark for instantaneous GL model selection that makes the following contributions. (i) GLEMOS provides extensive benchmark data for fundamental GL tasks, i.e., link prediction and node classification, including the performances of 366 models on 457 graphs on these tasks. (ii) GLEMOS designs multiple evaluation settings, and assesses how effectively representative model selection techniques perform in these different settings. (iii) GLEMOS is designed to be easily extended with new models, new graphs, and new performance records. (iv) Based on the experimental results, we discuss the limitations of existing approaches and highlight future research directions. To promote research on this significant problem, we make the benchmark data and code publicly available at https://github.com/facebookresearch/glemos., Comment: NeurIPS 2023

We calculate directly in position space the one-loop renormalization kernels of the soft operators $O_\gamma$ and $O_g$ that appear in the soft-quark contributions to, respectively, the subleading-power $\gamma\gamma\to h$ and $gg\to h$ form factors mediated by the $b$-quark. We present an IR/rapidity divergence-free definition for $O_g$ and demonstrate that with a correspondent definition of the collinear function, a consistent factorization theorem is recovered. Using conformal symmetry techniques, we establish a relation between the evolution kernels of the leading-twist heavy-light light-ray operator, whose matrix element defines the $B$-meson light-cone distribution amplitude (LCDA), and $O_\gamma$ to all orders in perturbation theory. Application of this relation allows us to bootstrap the kernel of $O_\gamma$ to the two-loop level. We construct an ansatz for the kernel of $O_g$ at higher orders. We test this ansatz against the consistency requirement at two-loop and find they differ only by a particular constant., Comment: journal version

Decision-making, a complicated task requiring various types of abilities, presents an excellent framework for assessing Large Language Models (LLMs). Our research investigates decision-making capabilities of LLMs through the lens of Game Theory. We focus specifically on games that support the simultaneous participation of more than two agents. We introduce GAMA($\gamma$)-Bench, which evaluates LLMs' Gaming Ability in Multi-Agent environments. $\gamma$-Bench includes eight classical multi-agent games and a scoring scheme specially designed to quantitatively assess LLMs' performance. Leveraging $\gamma$-Bench, we investigate LLMs' robustness, generalizability, and strategies for enhancement. Results reveal that while GPT-3.5 shows satisfying robustness, its generalizability is relatively limited. However, its performance can be improved through approaches such as Chain-of-Thought. Additionally, we evaluate twelve versions from six models, including GPT-3.5, GPT-4, Gemini, LLaMA-3.1, Mixtral, and Qwen-2. We find that Gemini-1.5-Pro outperforms other models with a score of $63.8$ out of $100$, followed by LLaMA-3.1-70B and GPT-4 with scores of $60.9$ and $60.5$, respectively. The code and experimental results are made publicly available via https://github.com/CUHK-ARISE/GAMABench., Comment: 11 pages of main text. 20 pages of appendices. 12 figures, 9 tables. Added models: Gemini-1.5-Pro, LLaMA-3.1-{7, 70, 405}B, Mixtral-8x{7, 22}B, Qwen-2-72B

Image restoration is a critical task in low-level computer vision, aiming to restore high-quality images from degraded inputs. Various models, such as convolutional neural networks (CNNs), generative adversarial networks (GANs), transformers, and diffusion models (DMs), have been employed to address this problem with significant impact. However, CNNs have limitations in capturing long-range dependencies. DMs require large prior models and computationally intensive denoising steps. Transformers have powerful modeling capabilities but face challenges due to quadratic complexity with input image size. To address these challenges, we propose VmambaIR, which introduces State Space Models (SSMs) with linear complexity into comprehensive image restoration tasks. We utilize a Unet architecture to stack our proposed Omni Selective Scan (OSS) blocks, consisting of an OSS module and an Efficient Feed-Forward Network (EFFN). Our proposed omni selective scan mechanism overcomes the unidirectional modeling limitation of SSMs by efficiently modeling image information flows in all six directions. Furthermore, we conducted a comprehensive evaluation of our VmambaIR across multiple image restoration tasks, including image deraining, single image super-resolution, and real-world image super-resolution. Extensive experimental results demonstrate that our proposed VmambaIR achieves state-of-the-art (SOTA) performance with much fewer computational resources and parameters. Our research highlights the potential of state space models as promising alternatives to the transformer and CNN architectures in serving as foundational frameworks for next-generation low-level visual tasks., Comment: 23 pages

Graph neural networks (GNNs) have achieved remarkable success across a wide range of applications, such as recommendation, drug discovery, and question answering. Behind the success of GNNs lies the backpropagation (BP) algorithm, which is the de facto standard for training deep neural networks (NNs). However, despite its effectiveness, BP imposes several constraints, which are not only biologically implausible, but also limit the scalability, parallelism, and flexibility in learning NNs. Examples of such constraints include storage of neural activities computed in the forward pass for use in the subsequent backward pass, and the dependence of parameter updates on non-local signals. To address these limitations, the forward-forward algorithm (FF) was recently proposed as an alternative to BP in the image classification domain, which trains NNs by performing two forward passes over positive and negative data. Inspired by this advance, we propose ForwardGNN in this work, a new forward learning procedure for GNNs, which avoids the constraints imposed by BP via an effective layer-wise local forward training. ForwardGNN extends the original FF to deal with graph data and GNNs, and makes it possible to operate without generating negative inputs (hence no longer forward-forward). Further, ForwardGNN enables each layer to learn from both the bottom-up and top-down signals without relying on the backpropagation of errors. Extensive experiments on real-world datasets show the effectiveness and generality of the proposed forward graph learning framework. We release our code at https://github.com/facebookresearch/forwardgnn., Comment: ICLR 2024

Transformer has become one of the most popular architectures for multivariate time series (MTS) forecasting. Recent Transformer-based MTS models generally prefer channel-independent structures with the observation that channel independence can alleviate noise and distribution drift issues, leading to more robustness. Nevertheless, it is essential to note that channel dependency remains an inherent characteristic of MTS, carrying valuable information. Designing a model that incorporates merits of both channel-independent and channel-mixing structures is a key to further improvement of MTS forecasting, which poses a challenging conundrum. To address the problem, an injection method for global information into channel-independent Transformer, InjectTST, is proposed in this paper. Instead of designing a channel-mixing model directly, we retain the channel-independent backbone and gradually inject global information into individual channels in a selective way. A channel identifier, a global mixing module and a self-contextual attention module are devised in InjectTST. The channel identifier can help Transformer distinguish channels for better representation. The global mixing module produces cross-channel global information. Through the self-contextual attention module, the independent channels can selectively concentrate on useful global information without robustness degradation, and channel mixing is achieved implicitly. Experiments indicate that InjectTST can achieve stable improvement compared with state-of-the-art models.

Large-scale high-resolution land cover classification is a prerequisite for constructing Earth system models and addressing ecological and resource issues. Advancements in satellite sensor technology have led to an improvement in spatial resolution and wider coverage areas. Nevertheless, the lack of high-resolution labeled data is still a challenge, hindering the largescale application of land cover classification methods. In this paper, we propose a Transformerbased weakly supervised method for cross-resolution land cover classification using outdated data. First, to capture long-range dependencies without missing the fine-grained details of objects, we propose a U-Net-like Transformer based on a reverse difference mechanism (RDM) using dynamic sparse attention. Second, we propose an anti-noise loss calculation (ANLC) module based on optimal transport (OT). Anti-noise loss calculation identifies confident areas (CA) and vague areas (VA) based on the OT matrix, which relieves the impact of noises in outdated land cover products. By introducing a weakly supervised loss with weights and employing unsupervised loss, the RDM-based U-Net-like Transformer was trained. Remote sensing images with 1 m resolution and the corresponding ground-truths of six states in the United States were employed to validate the performance of the proposed method. The experiments utilized outdated land cover products with 30 m resolution from 2013 as training labels, and produced land cover maps with 1 m resolution from 2017. The results show the superiority of the proposed method compared to state-of-the-art methods. The code is available at https://github.com/yu-ni1989/ANLC-Former.

Text watermarking technology aims to tag and identify content produced by large language models (LLMs) to prevent misuse. In this study, we introduce the concept of cross-lingual consistency in text watermarking, which assesses the ability of text watermarks to maintain their effectiveness after being translated into other languages. Preliminary empirical results from two LLMs and three watermarking methods reveal that current text watermarking technologies lack consistency when texts are translated into various languages. Based on this observation, we propose a Cross-lingual Watermark Removal Attack (CWRA) to bypass watermarking by first obtaining a response from an LLM in a pivot language, which is then translated into the target language. CWRA can effectively remove watermarks, decreasing the AUCs to a random-guessing level without performance loss. Furthermore, we analyze two key factors that contribute to the cross-lingual consistency in text watermarking and propose X-SIR as a defense method against CWRA. Code: https://github.com/zwhe99/X-SIR., Comment: ACL 2024 (main conference)

With the exponential growth of the life science literature, biomedical text mining (BTM) has become an essential technology for accelerating the extraction of insights from publications. Identifying named entities (e.g., diseases, drugs, or genes) in texts and their linkage to reference knowledge bases are crucial steps in BTM pipelines to enable information aggregation from different documents. However, tools for these two steps are rarely applied in the same context in which they were developed. Instead, they are applied in the wild, i.e., on application-dependent text collections different from those used for the tools' training, varying, e.g., in focus, genre, style, and text type. This raises the question of whether the reported performance of BTM tools can be trusted for downstream applications. Here, we report on the results of a carefully designed cross-corpus benchmark for named entity extraction, where tools were applied systematically to corpora not used during their training. Based on a survey of 28 published systems, we selected five for an in-depth analysis on three publicly available corpora encompassing four different entity types. Comparison between tools results in a mixed picture and shows that, in a cross-corpus setting, the performance is significantly lower than the one reported in an in-corpus setting. HunFlair2 showed the best performance on average, being closely followed by PubTator. Our results indicate that users of BTM tools should expect diminishing performances when applying them in the wild compared to original publications and show that further research is necessary to make BTM tools more robust.

Motivated by the success of unsupervised neural machine translation (UNMT), we introduce an unsupervised sign language translation and generation network (USLNet), which learns from abundant single-modality (text and video) data without parallel sign language data. USLNet comprises two main components: single-modality reconstruction modules (text and video) that rebuild the input from its noisy version in the same modality and cross-modality back-translation modules (text-video-text and video-text-video) that reconstruct the input from its noisy version in the different modality using back-translation procedure.Unlike the single-modality back-translation procedure in text-based UNMT, USLNet faces the cross-modality discrepancy in feature representation, in which the length and the feature dimension mismatch between text and video sequences. We propose a sliding window method to address the issues of aligning variable-length text with video sequences. To our knowledge, USLNet is the first unsupervised sign language translation and generation model capable of generating both natural language text and sign language video in a unified manner. Experimental results on the BBC-Oxford Sign Language dataset (BOBSL) and Open-Domain American Sign Language dataset (OpenASL) reveal that USLNet achieves competitive results compared to supervised baseline models, indicating its effectiveness in sign language translation and generation.

Insufficient modeling of human preferences within the reward model is a major obstacle for leveraging human feedback to improve translation quality. Fortunately, quality estimation (QE), which predicts the quality of a given translation without reference, has achieved impressive alignment with human evaluations in the last two years. In this work, we investigate the potential of employing the QE model as the reward model to predict human preferences for feedback training. We first identify the overoptimization problem during QE-based feedback training, manifested as an increase in reward while translation quality declines. We examine the problem and argue that the vulnerability of the QE model might lead to high rewards for incorrect translations, resulting in overoptimization and error propagation. To address the problem, we adopt a simple yet effective method that uses heuristic rules to detect the incorrect translations and assigns a penalty term to the reward scores of them. Experimental results show that the proposed QE-based feedback training achieves consistent and significant improvements across various settings, further verified through human preference studies. Our subsequent analysis demonstrates the high data efficiency of the proposed QE-based feedback training: it outperforms systems using larger parallel corpora by a small amount of monolingual data. Our code is available at: https://github.com/zwhe99/FeedbackMT, Comment: NAACL 2024

Recently, the impressive performance of large language models (LLMs) on a wide range of tasks has attracted an increasing number of attempts to apply LLMs in drug discovery. However, molecule optimization, a critical task in the drug discovery pipeline, is currently an area that has seen little involvement from LLMs. Most of existing approaches focus solely on capturing the underlying patterns in chemical structures provided by the data, without taking advantage of expert feedback. These non-interactive approaches overlook the fact that the drug discovery process is actually one that requires the integration of expert experience and iterative refinement. To address this gap, we propose DrugAssist, an interactive molecule optimization model which performs optimization through human-machine dialogue by leveraging LLM's strong interactivity and generalizability. DrugAssist has achieved leading results in both single and multiple property optimization, simultaneously showcasing immense potential in transferability and iterative optimization. In addition, we publicly release a large instruction-based dataset called MolOpt-Instructions for fine-tuning language models on molecule optimization tasks. We have made our code and data publicly available at https://github.com/blazerye/DrugAssist, which we hope to pave the way for future research in LLMs' application for drug discovery., Comment: Geyan Ye and Xibao Cai are equal contributors; Longyue Wang is corresponding author

This talk reviews recent developments in the field of analytical Feynman integral calculations. The central theme is the geometry associated to a given Feynman integral. In the simplest case this is a complex curve of genus zero (aka the Riemann sphere). In this talk we discuss Feynman integrals related to more complicated geometries like curves of higher genus or manifolds of higher dimensions. In the latter case we encounter Calabi-Yau manifolds. We also discuss how to compute these Feynman integrals., Comment: 11 pages, talk given at the conference Matter to the Deepest 2023. arXiv admin note: substantial text overlap with arXiv:2309.07531

Building on work by Hang and He, we show how the residual five-dimensional diffeomorphism symmetries of compactified gravitational theories with a warped extra dimension imply Equivalence theorems which ensure that the scattering amplitudes of helicity-0 and helicity-1 spin-2 Kaluza-Klein states equal (to leading order in scattering energy) those of the corresponding Goldstone bosons present in the `t-Hooft-Feynman gauge. We derive a set of Ward identities that lead to a transparent power-counting of the scattering amplitudes involving spin-2 Kaluza-Klein states. We explicitly calculate these amplitudes in terms of the Goldstone bosons in the Randall-Sundrum model, check the correspondence to previous unitary-gauge computations, and demonstrate the efficacy of `t-Hooft-Feynman gauge for accurately computing amplitudes for scattering of the spin-2 states both among themselves and with matter. Power-counting for the Goldstone boson interactions establishes that the scattering amplitudes grow no faster than $O(s)$, explaining the origin of the behavior previously shown to arise from intricate cancellations between different contributions to these scattering amplitudes in unitary gauge. We describe how our results apply to more general warped geometries, including models with a stabilized extra dimension. In an appendix we explicitly identify the symmetry algebra of the residual 5D diffeomorphisms of a Randall-Sundrum extra-dimensional theory., Comment: 19 pages, 7 figures

It has long been recognized that the scattering of electroweak particles at very high energies is dominated by vector boson fusion, which probes the origin of electroweak symmetry breaking and offers a unique window into the ultraviolet regime of the SM. Previous studies assume SM-like couplings and rely on the effective $W$ approximation (or electroweak parton distribution), whose validity is well-established within the SM but not yet studied in the presence of anomalous Higgs couplings. In this work, we critically examine the electroweak production of two Higgs bosons in the presence of anomalous $VVh$ and $VVhh$ couplings. We compute the corresponding helicity amplitudes and compare the cross section results in the effective $W$ approximation with the full fixed-order calculation. In particular, we identify two distinct classes of anomalous Higgs couplings, whose effects are not captured by vector boson fusion and effective $W$ approximation. Such very high energy electroweak scatterings can be probed at the Muon Shot, a multi-TeV muon collider upon which we base our study, although similar considerations apply to other high energy colliders., Comment: 33 pages, 13 figures, 3 tables