BZPEERReading, trust and collaboration for research
Menu

Discover

GraphFollow connections around a paper, topic or saved view.

Workspaces

HomePick up where your research flow left off.InboxHandle requests, replies and notifications from one place.CollectionsCurate reading lists, evidence sets and shareable dossiers.ResearchSee your private provenance graph, trail and imports.CollaborationMove from discovery into focused discussion rooms.PublishDraft, preview and publish full-text research articles.

Network

ExpertsFind specialists worth following or asking for help.CommunitiesJoin research circles organized around shared work.PartnersSee external organizations active around the same topics.

Opportunities

ListingsBrowse roles, calls and collaborations with clearer fit signals.

Account

Log inReturn to your reading and collaboration workspace.Create accountStart saving work, following people and joining teams.
Log inCreate account

Researcher profile

Yifei Zhang

Yifei Zhang contributes to research discovery and scholarly infrastructure.

ResearcherAffiliation not importedOpen to collaborate
Artificial IntelligenceComputation and LanguageComputer VisionMachine Learningastro-ph.IMphysics.app-phphysics.opticsastro-ph.HEnucl-exnucl-thphysics.ins-detastro-ph.COcond-mat.mtrl-scieess.SYhep-exSocial and Information NetworksSystems and Control

Trust snapshot

Quick read

Trust 21 - EmergingVerification L1Unclaimed author
31works
0followers
17topics
4close collaborators

Actions

Decide how to stay connected

Follow researcher0

Identity and collaboration

How to connect with this researcher

Claiming links this public author record to a researcher profile and unlocks direct collaboration workflows.

Log in to claim

Direct collaboration

Open a focused conversation when the fit is right

Claim this author entity first to unlock direct invitations.

Research graph

See the researcher in context

Open full explorer

Inspect adjacent work, topics, institutions and collaborators without jumping out to a separate graph page.

Building this graph slice

BZPEER is loading the nearby papers, people, topics and institutions for this page.

Published work

31 published item(s)

preprint2026arXiv

CIRAG: Construction-Integration Retrieval and Adaptive Generation for Multi-hop Question Answering

Triple-based Iterative Retrieval-Augmented Generation (iRAG) mitigates document-level noise for multi-hop question answering. However, existing methods still face limitations: (i) greedy single-path expansion, which propagates early errors and fails to capture parallel evidence from different reasoning branches, and (ii) granularity-demand mismatch, where a single evidence representation struggles to balance noise control with contextual sufficiency. In this paper, we propose the Construction-Integration Retrieval and Adaptive Generation model, CIRAG. It introduces an Iterative Construction-Integration module that constructs candidate triples and history-conditionally integrates them to distill core triples and generate the next-hop query. This module mitigates the greedy trap by preserving multiple plausible evidence chains. Besides, we propose an Adaptive Cascaded Multi-Granularity Generation module that progressively expands contextual evidence based on the problem requirements, from triples to supporting sentences and full passages. Moreover, we introduce Trajectory Distillation, which distills the teacher model's integration policy into a lightweight student, enabling efficient and reliable long-horizon reasoning. Extensive experiments demonstrate that CIRAG achieves superior performance compared to existing iRAG methods.

0 citations0 reviewsNo code linkedOpen access
preprint2026arXiv

CoSER: A Comprehensive Literary Dataset and Framework for Training and Evaluating LLM Role-Playing and Persona Simulation

Role-playing language agents (RPLAs) have emerged as promising applications of large language models (LLMs). However, simulating established characters presents a challenging task for RPLAs, due to the lack of authentic character datasets and nuanced evaluation methods using such data. In this paper, we present CoSER, a collection of a high-quality dataset, open models, and an evaluation protocol towards effective RPLAs of established characters. The CoSER dataset covers 17,966 characters from 771 renowned books. It provides authentic dialogues with real-world intricacies, as well as diverse data types such as conversation setups, character experiences and internal thoughts. Drawing from acting methodology, we introduce given-circumstance acting for training and evaluating role-playing LLMs, where LLMs sequentially portray multiple characters in book scenes. Using our dataset, we develop CoSER 8B and CoSER 70B, i.e., advanced open role-playing LLMs built on LLaMA-3.1 models. Extensive experiments demonstrate the value of the CoSER dataset for RPLA training, evaluation and retrieval. Moreover, CoSER 70B exhibits state-of-the-art performance surpassing or matching GPT-4o on our evaluation and three existing benchmarks, i.e., achieving 75.80% and 93.47% accuracy on the InCharacter and LifeChoice benchmarks respectively.

0 citations0 reviewsNo code linkedOpen access
preprint2026arXiv

From Concept to Capability: Evaluating 3D Gaussian Splatting for Synthetic Scene Editing in Autonomous Driving

The perception of an Autonomous Driving System (ADS) critically depends on relevant, comprehensive, and diverse datasets to ensure its safety while operating in the environment. Field data collection lacks completeness with respect to the list of rare but still possible safety-related scenarios needed for the development, verification, and validation of the ADS. 3D Gaussian Splatting (3DGS) has shown promising capabilities for the reconstruction and editing of scenes based on data collected by cameras and LiDAR sensors. However, the industrial fidelity evaluation of reconstructions is underexplored, which is crucial when employing such methods in safety-related systems, especially for ADS. This becomes more challenging as ADS operates in a dynamic, uncontrolled environment with limited viewpoints and often partially occluded objects. This paper addresses this gap by proposing and implementing a framework (Fig. 1) to systematically analyze the capabilities and limitations of 3DGS for use in the reconstruction of safety-related scenes. It focuses on the quality of reconstruction for vehicles and pedestrians, which are the two most critical object classes for ADS. Our findings provide industry insights into the fidelity degradation of reconstructions from multiple novel viewpoints, both lateral and longitudinal, enabling the integration of these methods into real-world industrial AD software development and testing pipelines.

0 citations0 reviewsNo code linkedOpen access
preprint2026arXiv

From Parameter Dynamics to Risk Scoring : Quantifying Sample-Level Safety Degradation in LLM Fine-tuning

Safety alignment of Large Language Models (LLMs) is extremely fragile, as fine-tuning on a small number of benign samples can erase safety behaviors learned from millions of preference examples. Existing studies attempt to explain this phenomenon by comparing parameters and hidden states before and after fine-tuning, but overlook their dynamic evolution during fine-tuning. In this paper, we uncover a critical mechanism underlying safety degradation by analyzing parameter dynamics, where benign fine-tuning causes parameters to cumulatively drift toward danger-aligned directions, progressively undermining the model's safety. This finding suggests that samples contributing more to this drift has greater fine-tuning risks. Based on this insight, we propose a method of Sample-Level Quantification of Safety Degradation (SQSD), which quantifies the influence of each training sample on safety degradation. Specifically, SQSD computes continuous risk scores to samples by measuring their induced parameter updates' projection difference between danger and safety directions. Extensive experiments across multiple models and datasets demonstrate that SQSD effectively quantifies sample-level fine-tuning risks and exhibits strong transferability across model architectures, parameter scales, and parameter-efficient methods.

0 citations0 reviewsNo code linkedOpen access
preprint2026arXiv

How Many Visual Tokens Do Multimodal Language Models Need? Scaling Visual Token Pruning with F^3A

Vision-language models improve perception by feeding increasingly long visual token sequences into language backbones, but the resulting inference cost raises a basic scaling question: as multimodal models grow, how many visual tokens are actually needed, and how should they be allocated under a fixed visual token budget? Existing training-free pruning methods typically answer this with one-shot proxies such as decoder attention, visual similarity, or conditional diversity. We argue that visual token pruning is better viewed as task-conditioned evidence search, especially under aggressive compression and across model scales. We propose F^3A, a training-free router for visual token pruning that operates before the language model consumes image tokens. F^3A builds lightweight question-conditioned cues, matches them to visual-grid tokens through frozen sparse sensing heads, and allocates a fixed vision token budget via coarse evidence localization, local refinement, coverage-preserving competition, and recovery of under-covered regions. It requires no model training, no extra LLM forward pass and preserves the original multimodal prompting and decoding pipeline.

0 citations0 reviewsNo code linkedOpen access
preprint2026arXiv

Masked Next-Scale Prediction for Self-supervised Scene Text Recognition

Scene Text Recognition requires modeling visual structures that evolve from coarse layouts to fine-grained character strokes. Training such models relies on large amounts of annotated data. Recent self-supervised approaches, such as Masked Image Modeling (MIM), alleviate this dependency by leveraging large-scale unlabeled data. Yet most existing MIM methods operate at a single spatial scale and fail to capture the hierarchical nature of scene text. In this work, we introduce Masked Next-Scale Prediction (MNSP), a unified self-supervised framework designed to explicitly model cross-scale structural evolution. The framework incorporates Next-Scale Prediction (NSP), which learns hierarchical representations by predicting higher-resolution features from lower-resolution contexts. Naive scale prediction, however, tends to produce spatially diffuse attention, directing the model toward background regions rather than textual structures. MNSP resolves this limitation by jointly learning cross-scale prediction and masked image reconstruction. NSP captures global layout priors across resolutions, while masked reconstruction imposes strong local constraints that guide attention toward informative text regions. A Multi-scale Linguistic Alignment module further maintains semantic consistency across different resolutions. Extensive experiments demonstrate that MNSP achieves state-of-the-art performance, reaching 86.2\% average accuracy on the challenging Union14M benchmark and 96.7\% across six standard datasets. Additional analyses show that our method improves robustness under extreme scale and layout variations. Code is available at https://github.com/CzhczhcHczh/MNSP

0 citations0 reviewsNo code linkedOpen access
preprint2026arXiv

MoLAN: A Unified Modality-Aware Noise Dynamic Editing Framework for Multimodal Sentiment Analysis

Multimodal Sentiment Analysis aims to integrate information from various modalities, such as audio, visual, and text, to make complementary predictions. However, it often struggles with irrelevant or misleading visual and auditory information. Most existing approaches typically treat the entire modality information (e.g., a whole image, audio segment, or text paragraph) as an independent unit for feature enhancement or denoising. They often suppress the redundant and noise information at the risk of losing critical information. To address this challenge, we propose MoLAN, a unified ModaLity-aware noise dynAmic editiNg framework. Specifically, MoLAN performs modality-aware blocking by dividing the features of each modality into multiple blocks. Each block is then dynamically assigned a distinct denoising strength based on its noise level and semantic relevance, enabling fine-grained noise suppression while preserving essential multimodal information. Notably, MoLAN is a unified and flexible framework that can be seamlessly integrated into a wide range of multimodal models. Building upon this framework, we further introduce MoLAN+, a new multimodal sentiment analysis approach. Experiments across five models and four datasets demonstrate the broad effectiveness of the MoLAN framework. Extensive evaluations show that MoLAN+ achieves the state-of-the-art performance. The code is publicly available at https://github.com/betterfly123/MoLAN-Framework.

0 citations0 reviewsNo code linkedOpen access
preprint2026arXiv

PlaM: Training-Free Plateau-Guided Model Merging for Better Visual Grounding in MLLMs

Multimodal Large Language Models (MLLMs) rely on strong linguistic reasoning inherited from their base language models. However, multimodal instruction fine-tuning paradoxically degrades this text's reasoning capability, undermining multimodal performance. To address this issue, we propose a training-free framework to mitigate this degradation. Through layer-wise vision token masking, we reveal a common three-stage pattern in multimodal large language models: early-modal separation, mid-modal alignment, and late-modal degradation. By analyzing the behavior of MLLMs at different stages, we propose a plateau-guided model merging method that selectively injects base language model parameters into MLLMs. Experimental results based on five MLLMs on nine benchmarks demonstrate the effectiveness of our method. Attention-based analysis further reveals that merging shifts attention from diffuse, scattered patterns to focused localization on task-relevant visual regions. Our repository is on https://github.com/wzj1718/PlaM.

0 citations0 reviewsNo code linkedOpen access
preprint2024arXiv

InRank: Incremental Low-Rank Learning

The theory of greedy low-rank learning (GLRL) aims to explain the impressive generalization capabilities of deep learning. It proves that stochastic gradient-based training implicitly regularizes neural networks towards low-rank solutions through a gradual increase of the rank during training. However, there is a gap between theory and practice since GLRL requires an infinitesimal initialization of the weights, which is not practical due to the fact that it is a saddle point. In this work, we remove the assumption of infinitesimal initialization by focusing on cumulative weight updates. We prove the cumulative weight updates follow an incremental low-rank trajectory for arbitrary orthogonal initialization of weights in a three-layer linear network. Empirically, we demonstrate that our theory holds on a broad range of neural networks (e.g., transformers) and standard training algorithms (e.g., SGD, Adam). However, existing training algorithms do not exploit the low-rank property to improve computational efficiency as the networks are not parameterized in low-rank. To remedy this, we design a new training algorithm Incremental Low-Rank Learning (InRank), which explicitly expresses cumulative weight updates as low-rank matrices while incrementally augmenting their ranks during training. We evaluate InRank on GPT-2, and our results indicate that InRank achieves comparable prediction performance as the full-rank counterpart while requiring at most 33% of the total ranks throughout training. We also propose an efficient version of InRank that achieves a reduction of 37% in total training time and 36% in model size when training GPT-medium on WikiText-103 from scratch.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Charm and beauty isolation from heavy flavor decay electrons in p+p and Pb+Pb collisions at $\sqrt{s_{\mathrm{NN}}}$ = 5.02 TeV at LHC

We present an analysis on the heavy flavor hadron decay electrons with charm and beauty contributions decomposed via a data driven method in p+p and Pb+Pb collisions at $\sqrt{s_{\mathrm{NN}}}$ = 5.02 TeV at LHC. The transverse momentum $p_{\mathrm{T}}$ spectra, nuclear modification factor $R_{\mathrm{AA}}$ and azimuthal anisotropic flow $v_2$ distributions of electrons from charm and beauty decays are obtained. We find that the electron $R_{\mathrm{AA}}$ from charm ($R_{\mathrm{AA}}^{\mathrm{c\rightarrow e}}$) and beauty ($R_{\mathrm{AA}}^{\mathrm{b\rightarrow e}}$) decays are suppressed at $p_{\mathrm{T}}$ $>$ 2.0 and $p_{\mathrm{T}}$ $>$ 3.0 GeV/$c$ in Pb+Pb collisions, respectively, which indicates that charm and beauty interact with and lose their energy in the hot-dense medium. A less suppression of electron $R_{\mathrm{AA}}$ from beauty decays than that from charm decays at 2.0 $<$ $p_{\mathrm{T}}$ $<$ 8.0 GeV/$c$ is observed, which is consistent with the mass-dependent partonic energy loss scenario. A non-zero electron $v_2$ from beauty decays ($v_{2}^{\mathrm{b\rightarrow e}}$) is observed and in good agreement with ALICE measurement. At low $p_{\mathrm{T}}$ region from 1.0 to 3.0 GeV/$c$, a discrepancy between RHIC and LHC results is observed with 68\% confidence level, which suggests different degree of thermalization of beauty quark under different temperatures of the medium. At 3.0 GeV/$c$ $<$ $p_{\mathrm{T}}$ $<$ 7.0 GeV/$c$, $v_{2}^{\mathrm{b\rightarrow e}}$ deviates from a number-of-constituent-quark (NCQ) scaling hypothesis, which favors that beauty quark is unlikely thermalized in heavy-ion collisions at LHC energy.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Electrical Programmable Multi-Level Non-volatile Photonic Random-Access Memory

Photonic Random-Access Memories (P-RAM) are an essential component for the on-chip non-von Neumann photonic computing by eliminating optoelectronic conversion losses in data links. Emerging Phase Change Materials (PCMs) have been showed multilevel memory capability, but demonstrations still yield relatively high optical loss and require cumbersome WRITE-ERASE approaches increasing power consumption and system package challenges. Here we demonstrate a multi-state electrically-programmed low-loss non-volatile photonic memory based on a broadband transparent phase change material (Ge2Sb2Se5, GSSe) with ultra-low absorption in the amorphous state. A zero-static-power and electrically-programmed multi-bit P-RAM is demonstrated on a silicon-on-insulator platform, featuring efficient amplitude modulation up to 0.2 dB/μm and an ultra-low insertion loss of total 0.12 dB for a 4-bit memory showing a 100x improved signal to loss ratio compared to other phase-change-materials based photonic memories. We further optimize the positioning of dual micro-heaters validating performance tradeoffs. Experimentally we demonstrate a half-a million cyclability test showcasing the robust approach of this material and device. Low-loss photonic retention-of-state adds a key feature for photonic functional and programmable circuits impacting many applications including neural networks, LiDAR, and sensors for example.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Graph-adaptive Rectified Linear Unit for Graph Neural Networks

Graph Neural Networks (GNNs) have achieved remarkable success by extending traditional convolution to learning on non-Euclidean data. The key to the GNNs is adopting the neural message-passing paradigm with two stages: aggregation and update. The current design of GNNs considers the topology information in the aggregation stage. However, in the updating stage, all nodes share the same updating function. The identical updating function treats each node embedding as i.i.d. random variables and thus ignores the implicit relationships between neighborhoods, which limits the capacity of the GNNs. The updating function is usually implemented with a linear transformation followed by a non-linear activation function. To make the updating function topology-aware, we inject the topological information into the non-linear activation function and propose Graph-adaptive Rectified Linear Unit (GReLU), which is a new parametric activation function incorporating the neighborhood information in a novel and efficient way. The parameters of GReLU are obtained from a hyperfunction based on both node features and the corresponding adjacent matrix. To reduce the risk of overfitting and the computational cost, we decompose the hyperfunction as two independent components for nodes and features respectively. We conduct comprehensive experiments to show that our plug-and-play GReLU method is efficient and effective given different GNN backbones and various downstream tasks.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

MulZDG: Multilingual Code-Switching Framework for Zero-shot Dialogue Generation

Building dialogue generation systems in a zero-shot scenario remains a huge challenge, since the typical zero-shot approaches in dialogue generation rely heavily on large-scale pre-trained language generation models such as GPT-3 and T5. The research on zero-shot dialogue generation without cumbersome language models is limited due to lacking corresponding parallel dialogue corpora. In this paper, we propose a simple but effective Multilingual learning framework for Zero-shot Dialogue Generation (dubbed as MulZDG) that can effectively transfer knowledge from an English corpus with large-scale training samples to a non-English corpus with zero samples. Besides, MulZDG can be viewed as a multilingual data augmentation method to improve the performance of the resource-rich language. First, we construct multilingual code-switching dialogue datasets via translation utterances randomly selected from monolingual English datasets. Then we employ MulZDG to train a unified multilingual dialogue model based on the code-switching datasets. The MulZDG can conduct implicit semantic alignment between different languages. Experiments on DailyDialog and DSTC7 datasets demonstrate that MulZDG not only achieve competitive performance under zero-shot case compared to training with sufficient examples but also greatly improve the performance of the source language.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Quasi-periodic oscillations of the X-ray burst from the magnetar SGR J1935+2154 and associated with the fast radio burst FRB 200428

The origin(s) and mechanism(s) of fast radio bursts (FRBs), which are short radio pulses from cosmological distances, have remained a major puzzle since their discovery. We report a strong Quasi-Periodic Oscillation(QPO) of 40 Hz in the X-ray burst from the magnetar SGR J1935+2154 and associated with FRB 200428, significantly detected with the Hard X-ray Modulation Telescope (Insight-HXMT) and also hinted by the Konus-Wind data. QPOs from magnetar bursts have only been rarely detected; our 3.4 sigma (p-value is 2.9e-4) detection of the QPO reported here reveals the strongest QPO signal observed from magnetars (except in some very rare giant flares), making this X-ray burst unique among magnetar bursts. The two X-ray spikes coinciding with the two FRB pulses are also among the peaks of the QPO. Our results suggest that at least some FRBs are related to strong oscillation processes of neutron stars. We also show that we may overestimate the significance of the QPO signal and underestimate the errors of QPO parameters if QPO exists only in a fraction of the time series of a X-ray burst which we use to calculate the Leahy-normalized periodogram.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Sample Complexity of the Robust LQG Regulator with Coprime Factors Uncertainty

This paper addresses the end-to-end sample complexity bound for learning the H2 optimal controller (the Linear Quadratic Gaussian (LQG) problem) with unknown dynamics, for potentially unstable Linear Time Invariant (LTI) systems. The robust LQG synthesis procedure is performed by considering bounded additive model uncertainty on the coprime factors of the plant. The closed-loop identification of the nominal model of the true plant is performed by constructing a Hankel-like matrix from a single time-series of noisy finite length input-output data, using the ordinary least squares algorithm from Sarkar et al. (2020). Next, an H-infinity bound on the estimated model error is provided and the robust controller is designed via convex optimization, much in the spirit of Boczar et al. (2018) and Zheng et al. (2020a), while allowing for bounded additive uncertainty on the coprime factors of the model. Our conclusions are consistent with previous results on learning the LQG and LQR controllers.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Significant Engagement Community Search on Temporal Networks: Concepts and Algorithms

Community search, retrieving the cohesive subgraph which contains the query vertex, has been widely touched over the past decades. The existing studies on community search mainly focus on static networks. However, real-world networks usually are temporal networks where each edge is associated with timestamps. The previous methods do not work when handling temporal networks. We study the problem of identifying the significant engagement community to which the user-specified query belongs. Specifically, given an integer k and a query vertex u, then we search for the subgraph H which satisfies (i) u $\in$ H; (ii) the de-temporal graph of H is a connected k-core; (iii) In H that u has the maximum engagement level. To address our problem, we first develop a top-down greedy peeling algorithm named TDGP, which iteratively removes the vertices with the maximum temporal degree. To boost the efficiency, we then design a bottom-up local search algorithm named BULS and its enhanced versions BULS+ and BULS*. Lastly, we empirically show the superiority of our proposed solutions on six real-world temporal graphs.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Statistical uncertainty estimation of higher-order cumulants with finite efficiency and its application in heavy-ion collisions

We derive the general analytical expressions for the statistical uncertainties of cumulants up to fourth order including an efficiency correction. The analytical expressions have been tested with a toy Monte Carlo model analysis. An application to the study of particle multiplicity fluctuations in heavy-ion collisions is investigated. In this derivation, a mathematical proof is given that the validity of an averaged efficiency correction and the fluctuations induced by the non-uniformity of efficiency can be eliminated. The estimation of statistical uncertainties using the analytical formulas is found to be significantly faster than the commonly used bootstrap method. The simplicity and efficiency of using the analytical formulas may be useful for massive data analysis in many fields.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

The removal method and generation mechanism of spikes in Insight-HXMT/HE telescope

Spikes are some obvious sharp increases that appear on the raw light curves of High Energy X-ray telescope(HE) onboard Insight-HXMT, which could have influences on the data products like energy and power spectra. They are considered to be fake triggers generated by large signals. In this paper, we study the characteristic of the spikes and propose two methods to remove spikes from the raw data. According to the different influences on energy and power spectra, the best parameters for removing the spikes is selected and used in the Insight-HXMT data analysis software. The generation mechanism of spikes is also studied using the backup HE detectors on ground and the spikes can be reduced by the electronic design.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Ultra-compact nonvolatile phase shifter based on electrically reprogrammable transparent phase change materials

Energy-efficient programmable photonic integrated circuits (PICs) are the cornerstone of on-chip classical and quantum optical technologies. Optical phase shifters constitute the fundamental building blocks which enable these programmable PICs. Thus far, carrier modulation and thermo-optical effect are the chosen phenomena for ultrafast and low-loss phase shifters, respectively; however, the state and information they carry are lost once the power is turned off-they are volatile. The volatility not only compromises energy efficiency due to their demand for constant power supply, but also precludes them from emerging applications such as in-memory computing. To circumvent this limitation, we introduce a novel phase shifting mechanism that exploits the nonvolatile refractive index modulation upon structural phase transition of Sb$_{2}$Se$_{3}$, a bi-stable transparent phase change material. A zero-static power and electrically-driven phase shifter was realized on a foundry-processed silicon-on-insulator platform, featuring record phase modulation up to 0.09 $π$/$μ$m and a low insertion loss of 0.3 dB/$π$, which can be further improved upon streamlined design. We also pioneered a one-step partial amorphization scheme to enhance the speed and energy efficiency of PCM devices. A diverse cohort of programmable photonic devices were demonstrated based on the ultra-compact PCM phase shifter.

0 citations0 reviewsNo code linkedOpen access
preprint2021arXiv

A Graph Reasoning Network for Multi-turn Response Selection via Customized Pre-training

We investigate response selection for multi-turn conversation in retrieval-based chatbots. Existing studies pay more attention to the matching between utterances and responses by calculating the matching score based on learned features, leading to insufficient model reasoning ability. In this paper, we propose a graph-reasoning network (GRN) to address the problem. GRN first conducts pre-training based on ALBERT using next utterance prediction and utterance order prediction tasks specifically devised for response selection. These two customized pre-training tasks can endow our model with the ability of capturing semantical and chronological dependency between utterances. We then fine-tune the model on an integrated network with sequence reasoning and graph reasoning structures. The sequence reasoning module conducts inference based on the highly summarized context vector of utterance-response pairs from the global perspective. The graph reasoning module conducts the reasoning on the utterance-level graph neural network from the local perspective. Experiments on two conversational reasoning datasets show that our model can dramatically outperform the strong baseline methods and can achieve performance which is close to human-level.

0 citations0 reviewsNo code linkedOpen access
preprint2021arXiv

In-orbit timing calibration of the Insight-Hard X-ray Modulation Telescope

We describe the timing system and the timing calibration results of the three payloads on-board the Insight-Hard X-ray Modulation Telescope (Insight-HXMT). These three payloads are the High Energy X-ray telescope (HE, 20-250 keV), the Medium Energy X-ray telescope (ME, 5-30 keV) and the low Energy X-ray telescope (LE, 1-10 keV). We present a method to correct the temperature-dependent period response and the long-term variation of the on-board crystal oscillator, especially for ME that does not carry a temperature-compensated crystal oscillator. The time of arrivals (ToAs) of the Crab pulsar are measured to evaluate the accuracy of the timing system. As the ephemeris of the Crab pulsar given by Jodrell Bank observatory has systematic errors around 40 μs (Rots et al. 2014), we use the quasi-simultaneous observations of the X-ray Timing Instrument (XTI) on-board the Neutron star Interior Composition Explorer (NICER) to produce the Crab ephemerides and to verify the timing system of Insight-HXMT. The energy-dependent ToAs&#39; offsets relative to the NICER measurements including physical and instrumental origins are about 24.7μs, 10.1μs and 864.7μs, and the systematic errors of the timing system are determined as 12.1μs, 8.6μs, and 15.8μs, for HE, ME and LE respectively.

0 citations0 reviewsNo code linkedOpen access
preprint2021arXiv

The design of the Ali CMB Polarization Telescope receiver

Ali CMB Polarization Telescope (AliCPT-1) is the first CMB degree-scale polarimeter to be deployed on the Tibetan plateau at 5,250m above sea level. AliCPT-1 is a 90/150 GHz 72 cm aperture, two-lens refracting telescope cooled down to 4 K. Alumina lenses, 800mm in diameter, image the CMB in a 33.4° field of view on a 636mm wide focal plane. The modularized focal plane consists of dichroic polarization-sensitive Transition-Edge Sensors (TESes). Each module includes 1,704 optically active TESes fabricated on a 150mm diameter silicon wafer. Each TES array is read out with a microwave multiplexing readout system capable of a multiplexing factor up to 2,048. Such a large multiplexing factor has allowed the practical deployment of tens of thousands of detectors, enabling the design of a receiver that can operate up to 19 TES arrays for a total of 32,376 TESes. AliCPT-1 leverages the technological advancements in the detector design from multiple generations of previously successful feedhorn-coupled polarimeters, and in the instrument design from BICEP-3, but applied on a larger scale. The cryostat receiver is currently under integration and testing. During the first deployment year, the focal plane will be populated with up to 4 TES arrays. Further TES arrays will be deployed in the following years, fully populating the focal plane with 19 arrays on the fourth deployment year. Here we present the AliCPT-1 receiver design, and how the design has been optimized to meet the experimental requirements.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

A Study of the Properties of the QCD Phase Diagram in High-Energy Nuclear Collisions

With the aim of understanding the phase structure of nuclear matter created in high-energy nuclear collisions at finite baryon density, a beam energy scan program has been carried out at Relativistic Heavy Ion Collider (RHIC). In this mini-review, most recent experimental results on collectivity, criticality and heavy flavor productions will be discussed. The goal here is to establish the connection between current available data and future heavy-ion collision experiments in a high baryon density region.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

Additively Homomorphical Encryption based Deep Neural Network for Asymmetrically Collaborative Machine Learning

The financial sector presents many opportunities to apply various machine learning techniques. Centralized machine learning creates a constraint which limits further applications in finance sectors. Data privacy is a fundamental challenge for a variety of finance and insurance applications that account on learning a model across different sections. In this paper, we define a new practical scheme of collaborative machine learning that one party owns data, but another party owns labels only, and term this \textbf{Asymmetrically Collaborative Machine Learning}. For this scheme, we propose a novel privacy-preserving architecture where two parties can collaboratively train a deep learning model efficiently while preserving the privacy of each party&#39;s data. More specifically, we decompose the forward propagation and backpropagation of the neural network into four different steps and propose a novel protocol to handle information leakage in these steps. Our extensive experiments on different datasets demonstrate not only stable training without accuracy loss, but also more than 100 times speedup compared with the state-of-the-art system.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

Design and Calibration of the High Energy Particle Monitor onboard the Insight-HXMT

Three high energy particle monitors (HPMs) employed onboard the Hard X-ray Modulation Telescope Insight-HXMT) can detect the charged particles from South Atlantic Anomaly (SAA) and hence provide the alert trigger for switch-on/off of the main detectors. Here a typical design of HPM with high stability and reliability is adopted by taking a plastic scintillator coupled with a small photomultiplier tube (PMT). The window threshold of HPM is designed as 1 MeV and 20 MeV for the incident electron and proton, respectively. Before the launch of Insight-HXMT, we performed in details the ground calibration of HPM. The measured energy response and its dependence on temperature are taken as essential input of Geant4 simulation for estimating the HPM count rate given with an incident particle energy spectrum. This serves as a guidance for choosing a reasonable working range of the PMT high voltage once the real SAA count rate is measured by HPM in orbit. So far the three HPMs have been working in orbit for more than two years. Apart from providing reliable alert trigger, the HPMs data are used as well to map the SAA region.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

In-flight calibration of the Insight-Hard X-ray Modulation Telescope

We present the calibration of the Insight-Hard X-ray Modulation Telescope (Insight-HXMT) X-ray satellite, which can be used to perform timing and spectral studies of bright X-ray sources. Insight-HXMT carries three main payloads onboard: the High Energy X-ray telescope (HE), the Medium Energy X-ray telescope (ME) and the Low Energy X-ray telescope (LE). In orbit, the radioactive sources, activated lines, the fluorescence lines and celestial sources are used to calibrate the energy scale and energy resolution of the payloads. The Crab nebular is adopted as the primary effective area calibrator and empirical functions are constructed to modify the simulated effective areas of the three payloads respectively. The systematic errors of HE, compared to the model of the Crab nebular, are less than 2% in 28--120 keV and 2%--10% above 120 keV. The systematic errors of ME are less than 1.5% in 10--35 keV. The systematic errors of LE are less than 1% in 1--7 keV except the Si K--edge (1.839 keV, up to 1.5%) and less than 2% in 7--10 keV.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

Multi-level Electro-thermal Switching of Optical Phase-Change Materials Using Graphene

Reconfigurable photonic systems featuring minimal power consumption are crucial for integrated optical devices in real-world technology. Current active devices available in foundries, however, use volatile methods to modulate light, requiring a constant supply of power and significant form factors. Essential aspects to overcoming these issues are the development of nonvolatile optical reconfiguration techniques which are compatible with on-chip integration with different photonic platforms and do not disrupt their optical performances. In this paper, a solution is demonstrated using an optoelectronic framework for nonvolatile tunable photonics that employs undoped-graphene microheaters to thermally and reversibly switch the optical phase-change material Ge$_2$Sb$_2$Se$_4$Te$_1$ (GSST). An in-situ Raman spectroscopy method is utilized to demonstrate, in real-time, reversible switching between four different levels of crystallinity. Moreover, a 3D computational model is developed to precisely interpret the switching characteristics, and to quantify the impact of current saturation on power dissipation, thermal diffusion, and switching speed. This model is used to inform the design of nonvolatile active photonic devices; namely, broadband Si$_3$N$_4$ integrated photonic circuits with small form-factor modulators and reconfigurable metasurfaces displaying 2$π$ phase coverage through neural-network-designed GSST meta-atoms. This framework will enable scalable, low-loss nonvolatile applications across a diverse range of photonics platforms.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

Progressive Cluster Purification for Unsupervised Feature Learning

In unsupervised feature learning, sample specificity based methods ignore the inter-class information, which deteriorates the discriminative capability of representation models. Clustering based methods are error-prone to explore the complete class boundary information due to the inevitable class inconsistent samples in each cluster. In this work, we propose a novel clustering based method, which, by iteratively excluding class inconsistent samples during progressive cluster formation, alleviates the impact of noise samples in a simple-yet-effective manner. Our approach, referred to as Progressive Cluster Purification (PCP), implements progressive clustering by gradually reducing the number of clusters during training, while the sizes of clusters continuously expand consistently with the growth of model representation capability. With a well-designed cluster purification mechanism, it further purifies clusters by filtering noise samples which facilitate the subsequent feature learning by utilizing the refined clusters as pseudo-labels. Experiments on commonly used benchmarks demonstrate that the proposed PCP improves baseline method with significant margins. Our code will be available at https://github.com/zhangyifei0115/PCP.

0 citations0 reviewsNo code linkedOpen access
preprint2019arXiv

Extensive beam test study of prototype MRPCs for the T0 detector at the CSR external-target experiment

The CSR External-target Experiment (CEE) will be the first large-scale nuclear physics experiment device at the Cooling Storage Ring (CSR) of the Heavy-Ion Research Facility in Lanzhou (HIRFL) in China. A new T0 detector has been proposed to measure the multiplicity, angular distribution and timing information of charged particles produced in heavy-ion collisions at the target region. Multi-gap resistive plate chamber (MRPC) technology was chosen as part of the construction of the T0 detector, which provides precision event collision times (T0) and collision geometry information. The prototype was tested with hadron and heavy-ion beams to study its performance. By comparing the experimental results with a Monte Carlo simulation, the time resolution of the MRPCs are found to be $\sim$ 50 ps or better. The timing performance of the T0 detector, including both detector and readout electronics, we found to fulfil the requirements of the CEE.

0 citations0 reviewsNo code linkedOpen access
preprint2019arXiv

Overview to the Hard X-ray Modulation Telescope (Insight-HXMT) Satellite

As China&#39;s first X-ray astronomical satellite, the Hard X-ray Modulation Telescope (HXMT), which was dubbed as Insight-HXMT after the launch on June 15, 2017, is a wide-band (1-250 keV) slat-collimator-based X-ray astronomy satellite with the capability of all-sky monitoring in 0.2-3 MeV. It was designed to perform pointing, scanning and gamma-ray burst (GRB) observations and, based on the Direct Demodulation Method (DDM), the image of the scanned sky region can be reconstructed. Here we give an overview of the mission and its progresses, including payload, core sciences, ground calibration/facility, ground segment, data archive, software, in-orbit performance, calibration, background model, observations and some preliminary results.

0 citations0 reviewsNo code linkedOpen access
preprint2019arXiv

Reconfigurable all-dielectric metalens with diffraction limited performance

Active metasurfaces, whose optical properties can be modulated post-fabrication, have emerged as an intensively explored field in recent years. The efforts to date, however, still face major performance limitations in tuning range, optical quality, and efficiency especially for non mechanical actuation mechanisms. In this paper, we introduce an active metasurface platform combining phase tuning covering the full 2$π$ range and diffraction-limited performance using an all-dielectric, low-loss architecture based on optical phase change materials (O-PCMs). We present a generic design principle enabling switching of metasurfaces between two arbitrary phase profiles and propose a new figure-of-merit (FOM) tailored for active meta-optics. We implement the approach to realize a high-performance varifocal metalens operating at 5.2 $μ$m wavelength. The metalens is constructed using Ge2Sb2Se4Te1 (GSST), an O-PCM with a large refractive index contrast ($Δ$ n > 1) and unique broadband low-loss characteristics in both amorphous and crystalline states. The reconfigurable metalens features focusing efficiencies above 20% at both states for linearly polarized light and a record large switching contrast ratio of 29.5 dB. We further validated aberration-free imaging using the metalens at both optical states, which represents the first experimental demonstration of a non-mechanical active metalens with diffraction-limited performance.

0 citations0 reviewsNo code linkedOpen access