Search

In this paper, we compute the total and differential cross sections for $e^+e^- \to J/\psi+c+\bar{c}$ at the $B$ factories up to next-to-leading order (NLO) corrections within the framework of nonrelativistic QCD factorization theory. We then obtain an improved pQCD prediction for this process by using the Principle of Maximum Conformality (PMC). We show the first time that the PMC can be applied for any pQCD calculable observable at the total or differential levels via a self-consistent way in perturbation theory. By the recursive use of renormalization group equation, the resultant PMC series satisfies all the requirements of renormalization group, which is free of conventional renormalization scheme-and-scale ambiguities at any fixed-order level. We thus obtain a more precise prompt total cross section at the NLO level, e.g. $\sigma|_{\rm prompt}^{\rm PMC}= 0.565^{+0.144}_{-0.125}~\text{pb}$. Here the uncertainty is the squared average of those from the $\alpha_s$ fixed-point uncertainty $\Delta\alpha_s(M_Z)$, the uncertainty of charm quark mass $\Delta m_c$, and an estimated contribution of the uncalculated NNLO-terms as predicted by the Pad\'{e} approximation approach. The differential cross sections $d\sigma/dP_{J/\psi}$, $d\sigma/d|\cos \theta|$, and $d\sigma/dz$ for $e^+e^- \to J/\psi+c+\bar{c}$ are further examined. Those results show that by further considering the feed-down contributions, the PMC predictions show better agreement with the Belle measurements., Comment: 22 pages, 11 figures

We present a calculation of the spin polarization of spin-1/2 fermions in a relativistic fluid at local thermodynamic equilibrium at the second order in the gradient expansion, including second-order derivatives. The second-order derivative terms vanish if the local equilibrium hypersurface is the hyperplane $t=const$ in the collision center-of-mass frame. However, since the freeze-out hypersurface has a non-trivial space-time structure, these terms may result in a non-vanishing contribution to the spin polarization, whose magnitude needs to be assessed with numerical computations., Comment: 18 pages, 2 figures

The chiral magnetic wave (CMW) is a collective mode in quark-gluon plasma originated from the chiral magnetic effect (CME) and chiral separation effect. Its detection in heavy-ion collisions is challenging due to significant background contamination. In Ref.[1], we have constructed a neural network which can accurately identify the CME-related signal from the final-state pion spectra. In this paper, we generalize such a neural network to the case of CMW search. We show that, after a updated training, the neural network can effectively recognize the CMW-related signal. Additionally, we assess the performance of the neural network compared to other known methods for CMW search., Comment: 6 pages, 6 figures. Published in Chin.Phys.C

In this paper, we explore the properties of the bottom-quark on-shell mass ($M_b$) by using its relation to the $\overline{\rm MS}$ mass (${\overline m}_b$). At present, this $\overline{\rm MS}$-on-shell relation has been known up to four-loop QCD corrections, which however still has a $\sim 2\%$ scale uncertainty by taking the renormalization scale as ${\overline m}_b({\overline m}_b)$ and varying it within the usual range of $[{\overline m}_b({\overline m}_b)/2, 2 {\overline m}_b({\overline m}_b)]$. The principle of maximum conformality (PMC) has been adopted to achieve a more precise $\overline{\rm MS}$-on-shell relation by eliminating such scale uncertainty. As a step forward, we also estimate the magnitude of the uncalculated higher-order terms by using the Pad\'{e} approximation approach. Numerically, by using the $\overline{\rm MS}$ mass ${\overline m}_b({\overline m}_b)=4.18^{+0.03}_{-0.02}$ GeV as an input, our predicted value for the bottom-quark on-shell mass becomes $M_b\simeq 5.36^{+0.10}_{-0.07}$ GeV, where the uncertainty is the squared average of the ones caused by $\Delta \alpha_s(M_Z)$, $\Delta {\overline m}_b({\overline m}_b)$, and the estimated magnitude of the higher-order terms., Comment: 5 pages, 2 figures

Split Neural Network, as one of the most common architectures used in vertical federated learning, is popular in industry due to its privacy-preserving characteristics. In this architecture, the party holding the labels seeks cooperation from other parties to improve model performance due to insufficient feature data. Each of these participants has a self-defined bottom model to learn hidden representations from its own feature data and uploads the embedding vectors to the top model held by the label holder for final predictions. This design allows participants to conduct joint training without directly exchanging data. However, existing research points out that malicious participants may still infer label information from the uploaded embeddings, leading to privacy leakage. In this paper, we first propose an embedding extension attack manipulating embeddings to undermine existing defense strategies, which rely on constraining the correlation between the embeddings uploaded by participants and the labels. Subsequently, we propose a new label obfuscation defense strategy, called `LabObf', which randomly maps each original integer-valued label to multiple real-valued soft labels with values intertwined, significantly increasing the difficulty for attackers to infer the labels. We conduct experiments on four different types of datasets, and the results show that LabObf significantly reduces the attacker's success rate compared to raw models while maintaining desirable model accuracy.

We record a particularly simple construction on top of Lumsdaine's local universes that allows for a Coquand-style universe of propositions with propositional extensionality to be interpreted in a category with subobject classifiers., Comment: 5 pages

Click-Through Rate (CTR) prediction holds a paramount position in recommender systems. The prevailing ID-based paradigm underperforms in cold-start scenarios due to the skewed distribution of feature frequency. Additionally, the utilization of a single modality fails to exploit the knowledge contained within textual features. Recent efforts have sought to mitigate these challenges by integrating Pre-trained Language Models (PLMs). They design hard prompts to structure raw features into text for each interaction and then apply PLMs for text processing. With external knowledge and reasoning capabilities, PLMs extract valuable information even in cases of sparse interactions. Nevertheless, compared to ID-based models, pure text modeling degrades the efficacy of collaborative filtering, as well as feature scalability and efficiency during both training and inference. To address these issues, we propose \textbf{C}ost-\textbf{E}fficient \textbf{L}anguage Model \textbf{A}lignment (\textbf{CELA}) for CTR prediction. CELA incorporates textual features and language models while preserving the collaborative filtering capabilities of ID-based models. This model-agnostic framework can be equipped with plug-and-play textual features, with item-level alignment enhancing the utilization of external information while maintaining training and inference efficiency. Through extensive offline experiments, CELA demonstrates superior performance compared to state-of-the-art methods. Furthermore, an online A/B test conducted on an industrial App recommender system showcases its practical effectiveness, solidifying the potential for real-world applications of CELA., Comment: 10 pages, 5 figures

Feature matching determines the orientation accuracy for the High Spatial Resolution (HSR) optical satellite stereos, subsequently impacting several significant applications such as 3D reconstruction and change detection. However, the matching of off-track HSR optical satellite stereos often encounters challenging conditions including wide-baseline observation, significant radiometric differences, multi-temporal changes, varying spatial resolutions, inconsistent spectral resolution, and diverse sensors. In this study, we evaluate various advanced feature matching algorithms for HSR optical satellite stereos. Utilizing a specially constructed dataset from five satellites across six challenging scenarios, HSROSS Dataset, we conduct a comparative analysis of four algorithms: the traditional SIFT, and deep-learning based methods including SuperPoint + SuperGlue, SuperPoint + LightGlue, and LoFTR. Our findings highlight overall superior performance of SuperPoint + LightGlue in balancing robustness, accuracy, distribution, and efficiency, showcasing its potential in complex HSR optical satellite scenarios., Comment: The manuscript is accepted as Oral Presentation in IEEE International Geoscience and Remote Sensing Symposium(IGARSS 2024)

Recently, large language models (LLMs) have demonstrated remarkable potential as an intelligent agent. However, existing researches mainly focus on enhancing the agent's reasoning or decision-making abilities through well-designed prompt engineering or task-specific fine-tuning, ignoring the procedure of exploration and exploitation. When addressing complex tasks within open-world interactive environments, these methods exhibit limitations. Firstly, the lack of global information of environments leads to greedy decisions, resulting in sub-optimal solutions. On the other hand, irrelevant information acquired from the environment not only adversely introduces noise, but also incurs additional cost. This paper proposes a novel approach, Weak Exploration to Strong Exploitation (WESE), to enhance LLM agents in solving open-world interactive tasks. Concretely, WESE involves decoupling the exploration and exploitation process, employing a cost-effective weak agent to perform exploration tasks for global knowledge. A knowledge graph-based strategy is then introduced to store the acquired knowledge and extract task-relevant knowledge, enhancing the stronger agent in success rate and efficiency for the exploitation task. Our approach is flexible enough to incorporate diverse tasks, and obtains significant improvements in both success rates and efficiency across four interactive benchmarks.

Motivated by the pair-density-wave (PDW) state found in the one-dimensional Kondo-Heisenberg chain, we report on a determinant quantum Monte Carlo DQMC study of pair-fields for a two-dimensional half-filled Hubbard layer coupled to an itinerant, non-interacting layer with one electron per site. In a specific range of interlayer hopping, the pairing vertex associated with PDW order becomes more attractive than that for uniform d-wave pairing, although both remain subdominant to the leading antiferromagnetic correlations at half-filling. Our result sheds light on where one potentially may find a PDW state in such a model.

Stance detection aims to determine the attitude expressed in text towards a given target. Zero-shot stance detection (ZSSD) has emerged to classify stances towards unseen targets during inference. Recent data augmentation techniques for ZSSD increase transferable knowledge between targets through text or target augmentation. However, these methods exhibit limitations. Target augmentation lacks logical connections between generated targets and source text, while text augmentation relies solely on training data, resulting in insufficient generalization. To address these issues, we propose an encoder-decoder data augmentation (EDDA) framework. The encoder leverages large language models and chain-of-thought prompting to summarize texts into target-specific if-then rationales, establishing logical relationships. The decoder generates new samples based on these expressions using a semantic correlation word replacement strategy to increase syntactic diversity. We also analyze the generated expressions to develop a rationale-enhanced network that fully utilizes the augmented data. Experiments on benchmark datasets demonstrate our approach substantially improves over state-of-the-art ZSSD techniques. The proposed EDDA framework increases semantic relevance and syntactic variety in augmented texts while enabling interpretable rationale-based learning.

This paper introduces RecAI, a practical toolkit designed to augment or even revolutionize recommender systems with the advanced capabilities of Large Language Models (LLMs). RecAI provides a suite of tools, including Recommender AI Agent, Recommendation-oriented Language Models, Knowledge Plugin, RecExplainer, and Evaluator, to facilitate the integration of LLMs into recommender systems from multifaceted perspectives. The new generation of recommender systems, empowered by LLMs, are expected to be more versatile, explainable, conversational, and controllable, paving the way for more intelligent and user-centric recommendation experiences. We hope the open-source of RecAI can help accelerate evolution of new advanced recommender systems. The source code of RecAI is available at \url{https://github.com/microsoft/RecAI}., Comment: 4 pages. Webconf 2024 demo track

As Large Language Models (LLMs) have shown significant intelligence, the progress to leverage LLMs as planning modules of autonomous agents has attracted more attention. This survey provides the first systematic view of LLM-based agents planning, covering recent works aiming to improve planning ability. We provide a taxonomy of existing works on LLM-Agent planning, which can be categorized into Task Decomposition, Plan Selection, External Module, Reflection and Memory. Comprehensive analyses are conducted for each direction, and further challenges for the field of research are discussed., Comment: 9 pages, 2 tables, 2 figures

This study investigates how Large Language Models (LLMs) leverage source and reference data in machine translation evaluation task, aiming to better understand the mechanisms behind their remarkable performance in this task. We design the controlled experiments across various input modes and model types, and employ both coarse-grained and fine-grained prompts to discern the utility of source versus reference information. We find that reference information significantly enhances the evaluation accuracy, while surprisingly, source information sometimes is counterproductive, indicating LLMs' inability to fully leverage the cross-lingual capability when evaluating translations. Further analysis of the fine-grained evaluation and fine-tuning experiments show similar results. These findings also suggest a potential research direction for LLMs that fully exploits the cross-lingual capability of LLMs to achieve better performance in machine translation evaluation tasks., Comment: Accepted by ACL2024 Findings

We investigate the effects of boost invariance breaking on the isotropization of pressure in the glasma, using the $3+1$D glasma simulation. The breaking is attributed to spatial fluctuations of the classical color charge density along the collision axis. We present numerical results for pressure and energy density at mid-rapidity and across a wider rapidity region. It is found that, despite varying longitudinal correlation lengths, the behaviors of the pressure isotropizations are qualitatively similar. The numerical results suggest that, in the initial stage, longitudinal color electromagnetic fields develop, similar to those in the boost invariant glasma. Subsequently, these fields evolve into a dilute glasma, expanding longitudinally in a manner akin to a dilute gas. We also show that the energy density at mid-rapidity exhibits a $1/\tau$ decay in the dilute glasma stage., Comment: 16 page, 6 figures

Cross-target stance detection (CTSD) is an important task, which infers the attitude of the destination target by utilizing annotated data derived from the source target. One important approach in CTSD is to extract domain-invariant features to bridge the knowledge gap between multiple targets. However, the analysis of informal and short text structure, and implicit expressions, complicate the extraction of domain-invariant knowledge. In this paper, we propose a Multi-Perspective Prompt-Tuning (MPPT) model for CTSD that uses the analysis perspective as a bridge to transfer knowledge. First, we develop a two-stage instruct-based chain-of-thought method (TsCoT) to elicit target analysis perspectives and provide natural language explanations (NLEs) from multiple viewpoints by formulating instructions based on large language model (LLM). Second, we propose a multi-perspective prompt-tuning framework (MultiPLN) to fuse the NLEs into the stance predictor. Extensive experiments results demonstrate the superiority of MPPT against the state-of-the-art baseline methods.