Search

Graph unlearning, which aims to eliminate the influence of specific nodes, edges, or attributes from a trained Graph Neural Network (GNN), is essential in applications where privacy, bias, or data obsolescence is a concern. However, existing graph unlearning techniques often necessitate additional training on the remaining data, leading to significant computational costs, particularly with large-scale graphs. To address these challenges, we propose a two-stage training-free approach, Erase then Rectify (ETR), designed for efficient and scalable graph unlearning while preserving the model utility. Specifically, we first build a theoretical foundation showing that masking parameters critical for unlearned samples enables effective unlearning. Building on this insight, the Erase stage strategically edits model parameters to eliminate the impact of unlearned samples and their propagated influence on intercorrelated nodes. To further ensure the GNN's utility, the Rectify stage devises a gradient approximation method to estimate the model's gradient on the remaining dataset, which is then used to enhance model performance. Overall, ETR achieves graph unlearning without additional training or full training data access, significantly reducing computational overhead and preserving data privacy. Extensive experiments on seven public datasets demonstrate the consistent superiority of ETR in model utility, unlearning efficiency, and unlearning effectiveness, establishing it as a promising solution for real-world graph unlearning challenges., Comment: Under review

Graph Neural Networks (GNNs) have demonstrated remarkable proficiency in handling a range of graph analytical tasks across various domains, such as e-commerce and social networks. Despite their versatility, GNNs face significant challenges in transferability, limiting their utility in real-world applications. Existing research in GNN transfer learning overlooks discrepancies in distribution among various graph datasets, facing challenges when transferring across different distributions. How to effectively adopt a well-trained GNN to new graphs with varying feature and structural distributions remains an under-explored problem. Taking inspiration from the success of Low-Rank Adaptation (LoRA) in adapting large language models to various domains, we propose GraphLoRA, an effective and parameter-efficient method for transferring well-trained GNNs to diverse graph domains. Specifically, we first propose a Structure-aware Maximum Mean Discrepancy (SMMD) to align divergent node feature distributions across source and target graphs. Moreover, we introduce low-rank adaptation by injecting a small trainable GNN alongside the pre-trained one, effectively bridging structural distribution gaps while mitigating the catastrophic forgetting. Additionally, a structure-aware regularization objective is proposed to enhance the adaptability of the pre-trained GNN to target graph with scarce supervision labels. Extensive experiments on six real-world datasets demonstrate the effectiveness of GraphLoRA against eleven baselines by tuning only 20% of parameters, even across disparate graph domains. The code is available at https://anonymous.4open.science/r/GraphLoRA., Comment: Under review

Reliable photo-nuclear reaction rates at the stellar conditions are essential to understand the origin of the heavy stable neutron-deficient isotopes between $^{74}$Se and $^{196}$Hg-p-nuclei, however, many reaction rates of relevance still have to rely on the Hauser-Feshbach model due to rare experimental progress. One such case is in the mass range of 160 for Dy, Er, Ho and Tm isotopes. In this work we attempt to constrain the Hauser-Feshbach model in the TALYS package by reproducing the available experimental data of $^{160}$Dy($p,\gamma$)$^{161}$Ho and $^{162}$Er($p,\gamma$)$^{163}$Tm in the $A\sim 160$ mass region, and examine the effects of level density, gamma strength function and the optical model potential. The constrained model then allows us to calculate the reaction rates of $^{157, 159}$Ho($\gamma$, $p$) and $^{163,165}$Tm($\gamma$, $p$) for the $\gamma$-process nucleosynthesis in carbon-deflagration SNe Ia model. Our recommended rates differ from the JINA REACLIB by more than 1 order of magnitude in the temperature range of 2-3 GK. This results in the changes of final abundance of $p$-nuclei in the $A\sim 160$ mass range by -5.5-3\% from those with JINA, which means that the ($\gamma$, $p$) reactions uncertainty is not predominant for the synthesis of these nuclei., Comment: 12 pages,7 figures

The use of smart devices as media for digital learning constitutes a new-generation digital learning paradigm. Therefore, context-aware game-based learning has attracted considerable attention. Location-based games have not only positive effects on learning but also pronounced effects on culture and history. Accordingly, focusing on railway cultural heritages, we attempted to assess interdependent relationships between key factors crucial for the design of a location-based mobile game for cultural heritages. We adopted the analytic network process (ANP) for our assessment. We initially performed a literature review to generalize relevant criteria and elements and developed a questionnaire based on the fuzzy delphi method (FDM); thus, key factors, namely 3 criteria and 15 elements, were selected. We also applied an online ANP-based questionnaire; on the basis of the experts opinions, we established a network model and determined the priority order of the key factors. The results revealed that experts considered culture learning to be of the highest importance, with the most important three elements being prior knowledge, challenge levels, and cultural narrative. In addition, culture learning exhibited a strong interaction with content design. In each criterion, one element had a considerable influence on the remaining elements, as determined by an analysis of matrices.

We investigate the behavior of geometric phase (GP) and geometric entanglement (GE), a multipartite entanglement measure, across quantum phase transitions in Rydberg atom chains. Using density matrix renormalization group calculations and finite-size scaling analysis, we characterize the critical properties of transitions between disordered and ordered phases. Both quantities exhibit characteristic scaling near transition points, with the disorder to $Z_2$ ordered phase transition showing behavior consistent with the Ising universality class, while the disorder to $Z_3$ phase transition displays distinct critical properties. We demonstrate that GP and GE serve as sensitive probes of quantum criticality, providing consistent critical parameters and scaling behavior. A unifying description of these geometric quantities from a quantum geometry perspective is explored, and an interferometric setup for their potential measurement is discussed. Our results provide insights into the interplay between geometric phase and multipartite entanglement near quantum phase transitions in Rydberg atom systems, revealing how these quantities reflect the underlying critical behavior in these complex quantum many-body systems.

Single metalenses are limited by their physical constraints, precluding themselves from achieving high numerical aperture across a wide visible spectral band in large-aperture applications. A hybrid system that integrates a metalens with a refractive lens can address this issue, yet previous designs lacked sufficient flexibility. Here, by reanalyzing the generalized Snell's law, we introduce a new paradigm for the hybrid metalens design based on differentiable ray tracing. Through joint optimization of the phase distribution of the metalens and refractive lens parameters, our system achieves achromatic performance within the broad spectral range of 440-700 nm, with an aperture of 1 cm and an f-number of 1.4. Owing to the differentiable nature of the proposed system, it can be seamlessly integrated as the optical front-end into any differentiable computational imaging system. Our system offers unprecedented opportunities for the advancement of metalenses in innovative optical design and computational imaging domains.

We investigate the Uhlmann phase, a generalization of the celebrated Berry phase, for Bose-Einstein condensates (BECs) at finite temperature. The Uhlmann phase characterizes topological properties of mixed states, in contrast to the Berry phase which is defined for pure states at zero temperature. Using the $SU(1,1)$ symmetry of the Bogoliubov Hamiltonian, we derive a general formula for the Uhlmann phase of BECs. Numerical calculations reveal that the Uhlmann phase can differ from the Berry phase in the zero-temperature limit, contrary to previous studies. As the temperature increases, the Uhlmann phase exhibits a winding behavior, and we relate the total winding degree to the Berry phase. This winding indicates that the Uhlmann phase takes values on a Riemann surface. Furthermore, we propose an experimental scheme to measure the Uhlmann phase of BECs by purifying the density matrix using an atomic interferometer., Comment: 11 pages,7 figures

Enhancing the attribution in large language models (LLMs) is a crucial task. One feasible approach is to enable LLMs to cite external sources that support their generations. However, existing datasets and evaluation methods in this domain still exhibit notable limitations. In this work, we formulate the task of attributed query-focused summarization (AQFS) and present WebCiteS, a Chinese dataset featuring 7k human-annotated summaries with citations. WebCiteS derives from real-world user queries and web search results, offering a valuable resource for model training and evaluation. Prior works in attribution evaluation do not differentiate between groundedness errors and citation errors. They also fall short in automatically verifying sentences that draw partial support from multiple sources. We tackle these issues by developing detailed metrics and enabling the automatic evaluator to decompose the sentences into sub-claims for fine-grained verification. Our comprehensive evaluation of both open-source and proprietary models on WebCiteS highlights the challenge LLMs face in correctly citing sources, underscoring the necessity for further improvement. The dataset and code will be open-sourced to facilitate further research in this crucial field., Comment: 20 pages, 7 figures, accepted to ACL 2024 main conference

To promote precision medicine, individualized treatment regimes (ITRs) are crucial for optimizing the expected clinical outcome based on patient-specific characteristics. However, existing ITR research has primarily focused on scenarios with categorical treatment options and a single outcome. In reality, clinicians often encounter scenarios with continuous treatment options and multiple, potentially competing outcomes, such as medicine efficacy and unavoidable toxicity. To balance these outcomes, a proper weight is necessary, which should be learned in a data-driven manner that considers both patient preference and clinician expertise. In this paper, we present a novel algorithm for developing individualized treatment regimes (ITRs) that incorporate continuous treatment options and multiple outcomes, utilizing observational data. Our approach assumes that clinicians are optimizing individualized patient utilities with sub-optimal treatment decisions that are at least better than random assignment. Treatment assignment is assumed to directly depend on the true underlying utility of the treatment rather than patient characteristics. The proposed method simultaneously estimates the weighting of composite outcomes and the decision-making process, allowing for construction of individualized treatment regimes with continuous doses. The proposed estimators can be used for inference and variable selection, facilitating the identification of informative treatment assignments and preference-associated variables. We evaluate the finite sample performance of our proposed method via simulation studies and apply it to a real data application of radiation oncology analysis., Comment: no

We report observations of unusual normal-state electronic conduction properties and superconducting characteristics of high-quality CoSi$_2$/Si films grown on silicon Si(100) and Si(111) substrates. A good understanding of these features shall help to address the underlying physics of the unconventional pairing symmetry recently observed in transparent CoSi$_2$/TiSi$_2$ heterojunctions [S. P. Chiu \textit{et al.}, Sci. Adv. \textbf{7}, eabg6569 (2021); Nanoscale \textbf{15}, 9179 (2023)], where CoSi$_2$/Si is a superconductor with a superconducting transition temperature $T_c \simeq$ (1.1--1.5) K, dependent on its dimensions, and TiSi$_2$ is a normal metal. In CoSi$_2$/Si films, we find a pronounced positive magnetoresistance caused by the weak-antilocalization effect, indicating a strong Rashba spin-orbit coupling (SOC). This SOC generates two-component superconductivity in CoSi$_2$/TiSi$_2$ heterojunctions. The CoSi$_2$/Si films are stable under ambient conditions and have ultralow 1/$f$ noise. Moreover, they can be patterned via the standard lithography techniques, which might be of considerable practical value for future scalable superconducting and quantum device fabrication., Comment: 15 pages, 12 figures, 2 tables

Next-basket recommendation (NBR) aims to infer the items in the next basket given the corresponding basket sequence. Existing NBR methods are mainly based on either message passing in a plain graph or transition modelling in a basket sequence. However, these methods only consider point-to-point binary item relations while item dependencies in real world scenarios are often in higher order. Additionally, the importance of the same item to different users varies due to variation of user preferences, and the relations between items usually involve various aspects. As pretrained language models (PLMs) excel in multiple tasks in natural language processing (NLP) and computer vision (CV), many researchers have made great efforts in utilizing PLMs to boost recommendation. However, existing PLM-based recommendation methods degrade when encountering Out-Of-Vocabulary (OOV) items. OOV items are those whose IDs are out of PLM's vocabulary and thus unintelligible to PLM. To settle the above challenges, we propose a novel method HEKP4NBR, which transforms the knowledge graph (KG) into prompts, namely Knowledge Tree Prompt (KTP), to help PLM encode the OOV item IDs in the user's basket sequence. A hypergraph convolutional module is designed to build a hypergraph based on item similarities measured by an MoE model from multiple aspects and then employ convolution on the hypergraph to model correlations among multiple items. Extensive experiments are conducted on HEKP4NBR on two datasets based on real company data and validate its effectiveness against multiple state-of-the-art methods.

The development of ultracold atom technology has enabled the precise investigations on quantum dynamics of quantum gases. Recently, inspired by experimental advancement, the $SU(1,1)$ echo, akin to the well-known $SU(2)$ spin echo, has been proposed for single-component Bose-Einstein condensate (BEC). In this paper, we investigate the possibility of quantum echo in the more intricate two-component BEC by fully exploiting its underlying symmetry, which is the Lie group $Sp(4,R)$. We demonstrate that quantum echo can occur for the two-component BEC by applying a driving protocol consisting of two steps in each period. The first step can be any Bogoliubov Hamiltonian, while the second step is a Hamiltonian with interactions turned off, which plays a similar role as the $\pi$-pulse in spin echo. We confirm our theoretical results with numerical calculations for different examples of two-component BEC. We further consider the effect of interactions between the excited boson modes on the quantum echo process and discuss the possible experiment implementation of this quantum echo., Comment: 11 pages,6 figures

Measuring physical observables requires averaging experimental outcomes over numerous identical measurements. The complete distribution function of possible outcomes or its Fourier transform, known as the full counting statistics, provides a more detailed description. This method captures the fundamental quantum fluctuations in many-body systems and has gained significant attention in quantum transport research. In this letter, we propose that cusp singularities in the full counting statistics are a novel tool for distinguishing between ordered and disordered phases. As a specific example, we focus on the superfluid-to-Mott transition in the Bose-Hubbard model and introduce $Z_A(\alpha)=\langle \exp({i\alpha \sum_{i\in A}(\hat{n}_i}-\overline{n}))\rangle $ with $\overline{n}=\langle n_i \rangle$. Through both analytical analysis and numerical simulations, we demonstrate that $\partial_\alpha \log Z_A(\alpha)$ exhibits a discontinuity near $\alpha=\pi$ in the superfluid phase when the subsystem size is sufficiently large, while it remains smooth in the Mott phase. This discontinuity can be interpreted as a first-order transition between different semi-classical configurations of vortices. We anticipate that our discoveries can be readily tested using state-of-the-art ultracold atom and superconducting qubit platforms., Comment: 7 pages, 4 figures

As the realm of spectral imaging applications extends its reach into the domains of mobile technology and augmented reality, the demands for compact yet high-fidelity systems become increasingly pronounced. Conventional methodologies, exemplified by coded aperture snapshot spectral imaging systems, are significantly limited by their cumbersome physical dimensions and form factors. To address this inherent challenge, diffractive optical elements (DOEs) have been repeatedly employed as a means to mitigate issues related to the bulky nature of these systems. Nonetheless, it's essential to note that the capabilities of DOEs primarily revolve around the modulation of the phase of light. Here, we introduce an end-to-end computational spectral imaging framework based on a polarization-multiplexed metalens. A distinguishing feature of this approach lies in its capacity to simultaneously modulate orthogonal polarization channels. When harnessed in conjunction with a neural network, it facilitates the attainment of high-fidelity spectral reconstruction. Importantly, the framework is intrinsically fully differentiable, a feature that permits the joint optimization of both the metalens structure and the parameters governing the neural network. The experimental results presented herein validate the exceptional spatial-spectral reconstruction performance, underscoring the efficacy of this system in practical, real-world scenarios. This innovative approach transcends the traditional boundaries separating hardware and software in the realm of computational imaging and holds the promise of substantially propelling the miniaturization of spectral imaging systems.

Recent advances in deep learning methods such as LLMs and Diffusion models have created a need for improved quantization methods that can meet the computational demands of these modern architectures while maintaining accuracy. Towards this goal, we study the advantages of FP8 data formats for post-training quantization across 75 unique network architectures covering a wide range of tasks, including machine translation, language modeling, text generation, image classification, generation, and segmentation. We examine three different FP8 representations (E5M2, E4M3, and E3M4) to study the effects of varying degrees of trade-off between dynamic range and precision on model accuracy. Based on our extensive study, we developed a quantization workflow that generalizes across different network architectures. Our empirical results show that FP8 formats outperform INT8 in multiple aspects, including workload coverage (92.64% vs. 65.87%), model accuracy and suitability for a broader range of operations. Furthermore, our findings suggest that E4M3 is better suited for NLP models, whereas E3M4 performs marginally better than E4M3 on computer vision tasks. The code is publicly available on Intel Neural Compressor: https://github.com/intel/neural-compressor.

Pre-trained language models (PLMs) have demonstrated strong performance in sequential recommendation (SR), which are utilized to extract general knowledge. However, existing methods still lack domain knowledge and struggle to capture users' fine-grained preferences. Meanwhile, many traditional SR methods improve this issue by integrating side information while suffering from information loss. To summarize, we believe that a good recommendation system should utilize both general and domain knowledge simultaneously. Therefore, we introduce an external knowledge base and propose Knowledge Prompt-tuning for Sequential Recommendation (\textbf{KP4SR}). Specifically, we construct a set of relationship templates and transform a structured knowledge graph (KG) into knowledge prompts to solve the problem of the semantic gap. However, knowledge prompts disrupt the original data structure and introduce a significant amount of noise. We further construct a knowledge tree and propose a knowledge tree mask, which restores the data structure in a mask matrix form, thus mitigating the noise problem. We evaluate KP4SR on three real-world datasets, and experimental results show that our approach outperforms state-of-the-art methods on multiple evaluation metrics. Specifically, compared with PLM-based methods, our method improves NDCG@5 and HR@5 by \textcolor{red}{40.65\%} and \textcolor{red}{36.42\%} on the books dataset, \textcolor{red}{11.17\%} and \textcolor{red}{11.47\%} on the music dataset, and \textcolor{red}{22.17\%} and \textcolor{red}{19.14\%} on the movies dataset, respectively. Our code is publicly available at the link: \href{https://github.com/zhaijianyang/KP4SR}{\textcolor{blue}{https://github.com/zhaijianyang/KP4SR}.}

Charge-changing cross section ($\sigma_{\text{cc}}$) measurements via the transmission method have made important progress recently aiming to determine the charge radii of exotic nuclei. In this work, we report a new $\sigma_{\text{cc}}$ measurement of 304(9) MeV/nucleon $^{28}$Si on carbon at the second Radioactive Ion Beam Line in Lanzhou (RIBLL2) and describe the data analysis procedure in detail. This procedure is essential to evaluate the systematic uncertainty in the transmission method. The determined $\sigma_{\mathrm{cc}}$ of 1125(11) mb is found to be consistent with the existing data at similar energies. The present work will serve as a reference in the $\sigma_{\text{cc}}$ determinations at RIBLL2., Comment: 9 pages, 13 figures, to be published in Chinese Physics C

Contrastive learning (CL) has become the de-facto learning paradigm in self-supervised learning on graphs, which generally follows the "augmenting-contrasting" learning scheme. However, we observe that unlike CL in computer vision domain, CL in graph domain performs decently even without augmentation. We conduct a systematic analysis of this phenomenon and argue that homophily, i.e., the principle that "like attracts like", plays a key role in the success of graph CL. Inspired to leverage this property explicitly, we propose HomoGCL, a model-agnostic framework to expand the positive set using neighbor nodes with neighbor-specific significances. Theoretically, HomoGCL introduces a stricter lower bound of the mutual information between raw node features and node embeddings in augmented views. Furthermore, HomoGCL can be combined with existing graph CL models in a plug-and-play way with light extra computational overhead. Extensive experiments demonstrate that HomoGCL yields multiple state-of-the-art results across six public datasets and consistently brings notable performance improvements when applied to various graph CL methods. Code is avilable at https://github.com/wenzhilics/HomoGCL., Comment: Accepted to KDD 2023 Research Track

Class imbalance is the phenomenon that some classes have much fewer instances than others, which is ubiquitous in real-world graph-structured scenarios. Recent studies find that off-the-shelf Graph Neural Networks (GNNs) would under-represent minor class samples. We investigate this phenomenon and discover that the subspaces of minor classes being squeezed by those of the major ones in the latent space is the main cause of this failure. We are naturally inspired to enlarge the decision boundaries of minor classes and propose a general framework GraphSHA by Synthesizing HArder minor samples. Furthermore, to avoid the enlarged minor boundary violating the subspaces of neighbor classes, we also propose a module called SemiMixup to transmit enlarged boundary information to the interior of the minor classes while blocking information propagation from minor classes to neighbor classes. Empirically, GraphSHA shows its effectiveness in enlarging the decision boundaries of minor classes, as it outperforms various baseline methods in class-imbalanced node classification with different GNN backbone encoders over seven public benchmark datasets. Code is avilable at https://github.com/wenzhilics/GraphSHA., Comment: Accepted to KDD 2023 Research Track

The bipartite graph structure has shown its promising ability in facilitating the subspace clustering and spectral clustering algorithms for large-scale datasets. To avoid the post-processing via k-means during the bipartite graph partitioning, the constrained Laplacian rank (CLR) is often utilized for constraining the number of connected components (i.e., clusters) in the bipartite graph, which, however, neglects the distribution (or normalization) of these connected components and may lead to imbalanced or even ill clusters. Despite the significant success of normalized cut (Ncut) in general graphs, it remains surprisingly an open problem how to enforce a one-step normalized cut for bipartite graphs, especially with linear-time complexity. In this paper, we first characterize a novel one-step bipartite graph cut (OBCut) criterion with normalized constraints, and theoretically prove its equivalence to a trace maximization problem. Then we extend this cut criterion to a scalable subspace clustering approach, where adaptive anchor learning, bipartite graph learning, and one-step normalized bipartite graph partitioning are simultaneously modeled in a unified objective function, and an alternating optimization algorithm is further designed to solve it in linear time. Experiments on a variety of general and large-scale datasets demonstrate the effectiveness and scalability of our approach.

When the Fermi Hubbard model was first introduced sixty years ago, one of the original motivations was to understand correlation effects in itinerant ferromagnetism. In the past two decades, ultracold Fermi gas in an optical lattice has been used to study the Fermi Hubbard model. However, the metallic ferromagnetic correlation was observed only in a recent experiment using frustrated lattices, and its underlying mechanism is not clear yet. In this letter, we point out that, under the particle--hole transformation, the single-particle ground state can exhibit double degeneracy in such a frustrated lattice. Therefore, the low-energy state exhibits valley degeneracy, reminiscent of multi-orbit physics in ferromagnetic transition metals. The local repulsive interaction leads to the valley Hund's rule, responsible for the observed ferromagnetism. We generalize this mechanism to distorted honeycomb lattices and square lattices with flux. This mechanism was first discussed by M\"uller-Hartmann in a simpler one-dimension model. However, this mechanism has not been widely discussed and has not been related to experimental observations before. Hence, our study not only explains the experimental findings but also enriches our understanding of itinerant ferromagnetism., Comment: 6 pages, 5 figures