Search

The widespread usage of online Large Language Models (LLMs) inference services has raised significant privacy concerns about the potential exposure of private information in user inputs to eavesdroppers or untrustworthy service providers. Existing privacy protection methods for LLMs suffer from insufficient privacy protection, performance degradation, or severe inference time overhead. In this paper, we propose PrivacyRestore to protect the privacy of user inputs during LLM inference. PrivacyRestore directly removes privacy spans in user inputs and restores privacy information via activation steering during inference. The privacy spans are encoded as restoration vectors. We propose Attention-aware Weighted Aggregation (AWA) which aggregates restoration vectors of all privacy spans in the input into a meta restoration vector. AWA not only ensures proper representation of all privacy spans but also prevents attackers from inferring the privacy spans from the meta restoration vector alone. This meta restoration vector, along with the query with privacy spans removed, is then sent to the server. The experimental results show that PrivacyRestore can protect private information while maintaining acceptable levels of performance and inference efficiency.

Jailbreaking attacks can enable Large Language Models (LLMs) to bypass the safeguard and generate harmful content. Existing jailbreaking defense methods have failed to address the fundamental issue that harmful knowledge resides within the model, leading to potential jailbreak risks for LLMs. In this paper, we propose a novel defense method called Eraser, which mainly includes three goals: unlearning harmful knowledge, retaining general knowledge, and maintaining safety alignment. The intuition is that if an LLM forgets the specific knowledge required to answer a harmful question, it will no longer have the ability to answer harmful questions. The training of Erase does not actually require the model's own harmful knowledge, and it can benefit from unlearning general answers related to harmful queries, which means it does not need assistance from the red team. The experimental results show that Eraser can significantly reduce the jailbreaking success rate for various attacks without compromising the general capabilities of the model. Our codes are available at https://github.com/ZeroNLP/Eraser.

Community search has been extensively studied in the past decades. In recent years, there is a growing interest in attributed community search that aims to identify a community based on both the query nodes and query attributes. A set of techniques have been investigated. Though the recent methods based on advanced learning models such as graph neural networks (GNNs) can achieve state-of-the-art performance in terms of accuracy, we notice that 1) they suffer from severe efficiency issues; 2) they directly model community search as a node classification problem and thus cannot make good use of interdependence among different entities in the graph. Motivated by these, in this paper, we propose a new neurAL attrIbuted Community sEarch model for large-scale graphs, termed ALICE. ALICE first extracts a candidate subgraph to reduce the search scope and subsequently predicts the community by the Consistency-aware Net , termed ConNet. Specifically, in the extraction phase, we introduce the density sketch modularity that uses a unified form to combine the strengths of two existing powerful modularities, i.e., classical modularity and density modularity. Based on the new modularity metric, we first adaptively obtain the candidate subgraph, formed by the k-hop neighbors of the query nodes, with the maximum modularity. Then, we construct a node-attribute bipartite graph to take attributes into consideration. After that, ConNet adopts a cross-attention encoder to encode the interaction between the query and the graph. The training of the model is guided by the structure-attribute consistency and the local consistency to achieve better performance. Extensive experiments over 11 real-world datasets including one billion-scale graph demonstrate the superiority of ALICE in terms of accuracy, efficiency, and scalability.

Community search has aroused widespread interest in the past decades. Among existing solutions, the learning-based models exhibit outstanding performance in terms of accuracy by leveraging labels to 1) train the model for community score learning, and 2) select the optimal threshold for community identification. However, labeled data are not always available in real-world scenarios. To address this notable limitation of learning-based models, we propose a pre-trained graph Transformer based community search framework that uses Zero label (i.e., unsupervised), termed TransZero. TransZero has two key phases, i.e., the offline pre-training phase and the online search phase. Specifically, in the offline pretraining phase, we design an efficient and effective community search graph transformer (CSGphormer) to learn node representation. To pre-train CSGphormer without the usage of labels, we introduce two self-supervised losses, i.e., personalization loss and link loss, motivated by the inherent uniqueness of node and graph topology, respectively. In the online search phase, with the representation learned by the pre-trained CSGphormer, we compute the community score without using labels by measuring the similarity of representations between the query nodes and the nodes in the graph. To free the framework from the usage of a label-based threshold, we define a new function named expected score gain to guide the community identification process. Furthermore, we propose two efficient and effective algorithms for the community identification process that run without the usage of labels. Extensive experiments over 10 public datasets illustrate the superior performance of TransZero regarding both accuracy and efficiency.

Controlling topological phases of light has allowed experimental observations of abundant topological phenomena and development of robust photonic devices. The prospect of more sophisticated controls with topological photonic devices for practical implementations requires high-level programmability. Here, we demonstrate a fully programmable topological photonic chip with large-scale integration of silicon photonic nanocircuits and microresonators. Photonic artificial atoms and their interactions in our compound system can be individually addressed and controlled, therefore allowing arbitrary altering of structural parameters and geometrical configurations for the observations of dynamic topological phase transitions and diverse photonic topological insulators. By individually programming artificial atoms on the generic chip, it has allowed comprehensive statistic characterisations of topological robustness against relatively weak disorders, as well as counterintuitive topological Anderson phase transitions induced by strong disorders. Our generic topological photonic chip that can be rapidly reprogrammed to implement multifunctionalities, prototypes a flexible and versatile platform for possible applications across fundamental science and topological technologies.

The superior performance of supervised classification methods in the information extraction (IE) area heavily relies on a large amount of gold standard data. Recent zero-shot classification methods converted the task to other NLP tasks (e.g., textual entailment) and used off-the-shelf models of these NLP tasks to directly perform inference on the test data without using a large amount of IE annotation data. A potentially valuable by-product of these methods is the large-scale silver standard data, i.e., pseudo-labeled data by the off-the-shelf models of other NLP tasks. However, there is no further investigation into the use of these data. In this paper, we propose a new framework, Clean-LaVe, which aims to utilize silver standard data to enhance the zero-shot performance. Clean-LaVe includes four phases: (1) Obtaining silver data; (2) Identifying relatively clean data from silver data; (3) Finetuning the off-the-shelf model using clean data; (4) Inference on the test data. The experimental results show that Clean-LaVe can outperform the baseline by 5% and 6% on TACRED and Wiki80 dataset in the zero-shot relation classification task, and by 3%-7% on Smile (Korean and Polish) in the zero-shot cross-lingual relation classification task, and by 8% on ACE05-E+ in the zero-shot event argument classification task. The code is share in https://github.com/wjw136/Clean_LaVe.git., Comment: accepted by coling2024. arXiv:2211.13883 is our first edition

Verifying entanglement between parties is essential for creating a secure quantum network, and Bell tests are the most rigorous method for doing so. However, if there is any signaling between the parties, then the violation of these inequalities can no longer be used to draw conclusions about the presence of entanglement. This is because signaling between the parties allows them to coordinate their measurement settings and outcomes, which can give rise to a violation of Bell inequalities even if the parties are not genuinely entangled. There is a pressing need to examine the role of signaling in quantum communication protocols from multiple perspectives, including communication security, physics foundations, and resource utilization while also promoting innovative technological applications. Here, we propose a semi-device independent protocol that allows us to numerically correct for effects of correlations in experimental probability distributions, caused by statistical fluctuations and experimental imperfections. Our noise robust protocol presents a relaxation of a tomography-based optimisation method called the steering robustness, that uses semidefinite programming to numerically identify the optimal quantum steering inequality without the need for resource-intensive tomography. The proposed protocol is numerically and experimentally analyzed in the context of random, misaligned measurements, correcting for signalling where necessary, resulting in a higher probability of violation compared to existing state-of-the-art inequalities. Our work demonstrates the power of semidefinite programming for entanglement verification and brings quantum networks closer to practical applications.

The frequency and orbital angular momentum (OAM) are independent physical properties of light that both offer unbounded degrees of freedom. However, creating, processing, and detecting high-dimensional OAM states have been a pivot and long-lasting task, as the complexity of the required optical systems scales up drastically with the OAM dimension. On the other hand, mature toolboxes -- such as optical frequency combs -- have been developed in the frequency domain for parallel measurements with excellent fidelity. Here we correlate the two dimensions into an equidistant comb structure on a photonic chip. Dissipative optical solitons formed in a nonlinear microresonator are emitted through the engraved angular gratings with each comb line carrying distinct OAM. Such one-to-one correspondence between the OAM and frequencies manifests state-of-the-art extinction ratios over 18.5 dB, enabling precision spectroscopy of optical vortices. The demonstrated vortex soliton microcombs provide coherent light sources that are multiplexed in the spatial and frequency domain, having the potential to establish a new modus operandi of high-dimensional structured light.

The superior performance of supervised relation extraction (RE) methods heavily relies on a large amount of gold standard data. Recent zero-shot relation extraction methods converted the RE task to other NLP tasks and used off-the-shelf models of these NLP tasks to directly perform inference on the test data without using a large amount of RE annotation data. A potentially valuable by-product of these methods is the large-scale silver standard data. However, there is no further investigation on the use of potentially valuable silver standard data. In this paper, we propose to first detect a small amount of clean data from silver standard data and then use the selected clean data to finetune the pretrained model. We then use the finetuned model to infer relation types. We also propose a class-aware clean data detection module to consider class information when selecting clean data. The experimental results show that our method can outperform the baseline by 12% and 11% on TACRED and Wiki80 dataset in the zero-shot RE task. By using extra silver standard data of different distributions, the performance can be further improved., Comment: Our research group conducted further research and discussions on the same topic and the latest paper is here arXiv:2402.18061, which has been accepted in LREC-COLING 2024

Simulating the dynamic evolutions of physical and molecular systems in a quantum computer is of fundamental interest in many applications. Its implementation requires efficient quantum simulation algorithms. The Lie-Trotter-Suzuki approximation algorithm, also well known as the Trotterization, is a basic algorithm in quantum dynamic simulation. A multi-product algorithm that is a linear combination of multiple Trotterizations has been proposed to improve the approximation accuracy. Implementing such multi-product Trotterization in quantum computers however remains experimentally challenging and its success probability is limited. Here, we modify the multi-product Trotterization and combine it with the oblivious amplitude amplification to simultaneously reach a high simulation precision and high success probability. We experimentally implement the modified multi-product algorithm in an integrated-photonics programmable quantum simulator in silicon, which allows the initialization, manipulation and measurement of four-qubit states and a sequence of linearly combined controlled-unitary gates, to emulate the dynamics of a coupled electron and nuclear spins system. Theoretical and experimental results are in good agreement, and they both show the modified multi-product algorithm can simulate Hamiltonian dynamics with a higher precision than conventional Trotterizations and a nearly deterministic success probability. We certificate the multi-product algorithm in a small-scale quantum simulator based on linear combinations of operations, and this work promises the practical implementations of quantum dynamics simulations., Comment: 14 pages, 5 figures

Characterization and categorization of quantum correlations are both fundamentally and practically important in quantum information science. Although quantum correlations such as non-separability, steerability, and non-locality can be characterized by different theoretical models in different scenarios with either known (trusted) or unknown (untrusted) knowledge of the associated systems, such characterization sometimes lacks unambiguous to experimentalist. In this work, we propose the physical interpretation of nonlocal quantum correlation between two systems. In the absence of {\it complete local description} of one of the subsystems quantified by the {\it local uncertainty relation}, the correlation between subsystems becomes nonlocal. Remarkably, different nonlocal quantum correlations can be discriminated from a single uncertainty relation derived under local hidden state (LHS)-LHS model only. We experimentally characterize the two-qubit Werner state in different scenarios., Comment: 13 pages, 10 figures. Comments are welcome

With high integration density and excellent optical properties, silicon photonics is becoming a promising platform for complete integration and large-scale optical quantum information processing. Scalable quantum information applications need photon generation and detection to be integrated on the same chip, and we have seen that various devices on the silicon photonic chip have been developed for this goal. This paper reviews the relevant research results and state-of-the-art technologies on the silicon photonic chip for scalable quantum applications. Despite the shortcomings, properties of some components have already met the requirements for further expansion. Furthermore, we point out the challenges ahead and further research directions for on-chip scalable quantum information applications.

Quantum technology is poised to enable a step change in human capability for computing, communications and sensing. Photons are indispensable as carriers of quantum information - they travel at the fastest possible speed and readily protected from decoherence. However, the system requires thousands of near-transparent components with ultra-low-latency control. For quantum technology to be implemented, a new paradigm photonic system is required: one with in-built coherence, stability, the ability to define arbitrary circuits, and a path to manufacturability. Silicon photonics has unparalleled density and component performance, which, with CMOS compatible fabrication, place it in a strong position for a scalable quantum photonics platform. This paper is a progress report on silicon quantum photonics, focused on developments in the past five years. We provide an introduction on silicon quantum photonic component and the challenges in the field, summarise the current state-of-the-art and identify outstanding technical challenges, as well as promising avenues of future research. We also resolve a conflict in the definition of Hong-Ou-Mandel interference visibility in integrated quantum photonic experiments, needed for fair comparison of photon quality across different platforms. Our aim is the development of scalability on the platform, to which end we point the way to ever-closer integration, toward silicon quantum photonic systems-on-a-chip.