arXiv subjects

Xin Liu

Publications and source records attributed to Xin Liu.

At least 631 records · Page 35Linked to original sources

arXiv2023

Unsupervised HDR Image and Video Tone Mapping via Contrastive Learning

Capturing high dynamic range (HDR) images (videos) is attractive because it can reveal the details in both dark and bright regions. Since the mainstream screens only support low dynamic range (LDR) content, tone mapping algorithm is required to compress the dynamic range of HDR images (videos). Although image tone mapping has been widely explored, video tone mapping is lagging behind, especially for the deep-learning-based methods, due to the lack of HDR-LDR video pairs. In this work, we propose a unified framework (IVTMNet) for unsupervised image and video tone mapping. To improve unsupervised training, we propose domain and instance based contrastive learning loss. Instead of using a universal feature extractor, such as VGG to extract the features for similarity measurement, we propose a novel latent code, which is an aggregation of the brightness and contrast of extracted features, to measure the similarity of different pairs. We totally construct two negative pairs and three positive pairs to constrain the latent codes of tone mapped results. For the network structure, we propose a spatial-feature-enhanced (SFE) module to enable information exchange and transformation of nonlocal regions. For video tone mapping, we propose a temporal-feature-replaced (TFR) module to efficiently utilize the temporal correlation and improve the temporal consistency of video tone-mapped results. We construct a large-scale unpaired HDR-LDR video dataset to facilitate the unsupervised training process for video tone mapping. Experimental results demonstrate that our method outperforms state-of-the-art image and video tone mapping methods. Our code and dataset are available at https://github.com/cao-cong/UnCLTMO.

cs.CV↗

arXiv2023

Massive Black Hole Binaries as LISA Precursors in the Roman High Latitude Time Domain Survey

With its capacity to observe $\sim 10^{5-6}$ faint active galactic nuclei (AGN) out to redshift $z\approx 6$, Roman is poised to reveal a population of $10^{4-6}\, {\rm M_\odot}$ black holes during an epoch of vigorous galaxy assembly. By measuring the light curves of a subset of these AGN and looking for periodicity, Roman can identify several hundred massive black hole binaries (MBHBs) with 5-12 day orbital periods, which emit copious gravitational radiation and will inevitably merge on timescales of $10^{3-5}$ years. During the last few months of their merger, such binaries are observable with the Laser Interferometer Space Antenna (LISA), a joint ESA/NASA gravitational wave mission set to launch in the mid-2030s. Roman can thus find LISA precursors, provide uniquely robust constraints on the LISA source population, help identify the host galaxies of LISA mergers, and unlock the potential of multi-messenger astrophysics with massive black hole binaries.

astro-ph.HE↗

arXiv2023

Varstrometry for Off-nucleus and Dual sub-Kpc AGN (VODKA): Investigating the Nature of J0823+2418 at $z=1.81$: a Likely Lensed Quasar

Dual quasars at small physical separations are an important precursor phase of galaxy mergers, ultimately leading to the coalescence of the two supermassive black holes. Starting from a sample of dual/lensed quasar candidates discovered using astrometric jitter in Gaia data, we present a pilot case study of one of the most promising yet puzzling candidate dual quasars at cosmic noon (z$\sim$1.8). Using multi-wavelength imaging and spectroscopy from X-ray to radio, we test whether the J0823+2418 system is two individual quasars in a bound pair at separation$\sim$0.64'', or instead a single quasar being gravitationally lensed by a foreground galaxy. We find consistent flux ratios ($\sim$1.25-1.45) between the two sources in optical, NIR, UV, and radio, and thus similar spectral energy distributions, suggesting a strong lensing scenario. However, differences in the radio spectral index, as well as changing X-ray fluxes, hint at either a dual quasar with otherwise nearly identical properties, or perhaps lensing-based time lag of $\sim$3 days paired with intrinsic variability. We find with lens mass modeling that the relative NIR positions and magnitudes of the two quasars and a marginally detected central galaxy are consistent with strong lensing. Archival SDSS spectra likewise suggest a foreground absorber via Mg II absorption lines. We conclude that J0823+2418 is likely a lensed quasar, and therefore that the VODKA sample contains a population of these lensed systems (perhaps as high as 50%) as well as dual quasars.

astro-ph.GA↗

arXiv2023

rPPG-MAE: Self-supervised Pre-training with Masked Autoencoders for Remote Physiological Measurement

Remote photoplethysmography (rPPG) is an important technique for perceiving human vital signs, which has received extensive attention. For a long time, researchers have focused on supervised methods that rely on large amounts of labeled data. These methods are limited by the requirement for large amounts of data and the difficulty of acquiring ground truth physiological signals. To address these issues, several self-supervised methods based on contrastive learning have been proposed. However, they focus on the contrastive learning between samples, which neglect the inherent self-similar prior in physiological signals and seem to have a limited ability to cope with noisy. In this paper, a linear self-supervised reconstruction task was designed for extracting the inherent self-similar prior in physiological signals. Besides, a specific noise-insensitive strategy was explored for reducing the interference of motion and illumination. The proposed framework in this paper, namely rPPG-MAE, demonstrates excellent performance even on the challenging VIPL-HR dataset. We also evaluate the proposed method on two public datasets, namely PURE and UBFC-rPPG. The results show that our method not only outperforms existing self-supervised methods but also exceeds the state-of-the-art (SOTA) supervised methods. One important observation is that the quality of the dataset seems more important than the size in self-supervised pre-training of rPPG. The source code is released at https://github.com/linuxsino/rPPG-MAE.

cs.CV↗

arXiv2023

Accurate Estimation of Transport Coefficients Using Model-free Time Correlation Functions in Equilibrium Simulations

Transport coefficients, such as the diffusion coefficient and shear viscosity, are important material properties that are calculated in computer simulations. In this study, the criterion for the best estimation of viscosity, as an example of transport coefficients, is determined by using the Green-Kubo formula without any artificial models. The related algorithm is given by the estimation of the viscosities of polyethylene oxide solutions by using a molecular dynamics simulation for testing. The algorithm can be used in the simulations of complex systems with a long tail of correlations typically found in macromolecular and biological simulation systems.

physics.comp-ph↗

arXiv2023

A Simple yet Effective Self-Debiasing Framework for Transformer Models

Current Transformer-based natural language understanding (NLU) models heavily rely on dataset biases, while failing to handle real-world out-of-distribution (OOD) instances. Many methods have been proposed to deal with this issue, but they ignore the fact that the features learned in different layers of Transformer-based NLU models are different. In this paper, we first conduct preliminary studies to obtain two conclusions: 1) both low- and high-layer sentence representations encode common biased features during training; 2) the low-layer sentence representations encode fewer unbiased features than the highlayer ones. Based on these conclusions, we propose a simple yet effective self-debiasing framework for Transformer-based NLU models. Concretely, we first stack a classifier on a selected low layer. Then, we introduce a residual connection that feeds the low-layer sentence representation to the top-layer classifier. In this way, the top-layer sentence representation will be trained to ignore the common biased features encoded by the low-layer sentence representation and focus on task-relevant unbiased features. During inference, we remove the residual connection and directly use the top-layer sentence representation to make predictions. Extensive experiments and indepth analyses on NLU tasks show that our framework performs better than several competitive baselines, achieving a new SOTA on all OOD test sets.

cs.CL↗

arXiv2023

Deterministic topological quantum gates for Majorana qubits without ancillary modes

The realization of quantum gates in topological quantum computation still confronts significant challenges in both fundamental and practical aspects. Here, we propose a deterministic and fully topologically protected measurement-based scheme to realize the issue of implementing Clifford quantum gates on the Majorana qubits. Our scheme is based on rigorous proof that the single-qubit gate can be performed by leveraging the neighboring Majorana qubit but not disturbing its carried quantum information, eliminating the need for ancillary Majorana zero modes (MZMs) in topological quantum computing. Benefiting from the ancilla-free construction, we show the minimum measurement sequences with four steps to achieve two-qubit Clifford gates by constructing their geometric visualization. To avoid the uncertainty of the measurement-only strategy, we propose manipulating the MZMs in their parameter space to correct the undesired measurement outcomes while maintaining complete topological protection, as demonstrated in a concrete Majorana platform. Our scheme identifies the minimal operations of measurement-based topological and deterministic Clifford gates and offers an ancilla-free design of topological quantum computation.

cond-mat.mes-hall↗

arXiv2023

Optimizing Airbnb Search Journey with Multi-task Learning

At Airbnb, an online marketplace for stays and experiences, guests often spend weeks exploring and comparing multiple items before making a final reservation request. Each reservation request may then potentially be rejected or cancelled by the host prior to check-in. The long and exploratory nature of the search journey, as well as the need to balance both guest and host preferences, present unique challenges for Airbnb search ranking. In this paper, we present Journey Ranker, a new multi-task deep learning model architecture that addresses these challenges. Journey Ranker leverages intermediate guest actions as milestones, both positive and negative, to better progress the guest towards a successful booking. It also uses contextual information such as guest state and search query to balance guest and host preferences. Its modular and extensible design, consisting of four modules with clear separation of concerns, allows for easy application to use cases beyond the Airbnb search ranking context. We conducted offline and online testing of the Journey Ranker and successfully deployed it in production to four different Airbnb products with significant business metrics improvements.

cs.IR↗

arXiv2023

Large Language Models are Few-Shot Health Learners

Large language models (LLMs) can capture rich representations of concepts that are useful for real-world tasks. However, language alone is limited. While existing LLMs excel at text-based inferences, health applications require that models be grounded in numerical data (e.g., vital signs, laboratory values in clinical domains; steps, movement in the wellness domain) that is not easily or readily expressed as text in existing training corpus. We demonstrate that with only few-shot tuning, a large language model is capable of grounding various physiological and behavioral time-series data and making meaningful inferences on numerous health tasks for both clinical and wellness contexts. Using data from wearable and medical sensor recordings, we evaluate these capabilities on the tasks of cardiac signal analysis, physical activity recognition, metabolic calculation (e.g., calories burned), and estimation of stress reports and mental health screeners.

cs.CL↗

arXiv2023

The Spitzer Coverage of HSC-Deep with IRAC for Z studies (SHIRAZ) I: IRAC mosaics

We present new Spitzer Infrared Array Camera (IRAC) 3.6 and 4.5μm mosaics of three fields, E-COSMOS, DEEP2-F3, and ELAIS-N1. Our mosaics include both new IRAC observations as well as re-processed archival data in these fields. These fields are part of the HSC-Deep grizy survey and have a wealth of additional ancillary data. The addition of these new IRAC mosaics is critical in allowing for improved photometric redshifts and stellar population parameters at cosmic noon and earlier epochs. The total area mapped by this work is {\sim} 17 deg2 with a mean integration time of {\sim}1200s, providing a median 5σ depth of 23.7(23.3) at 3.6(4.5)μm in AB. We perform SExtractor photometry both on the combined mosaics as well as the single-epoch mosaics taken {\sim}6 months apart. The resultant IRAC number counts show good agreement with previous studies. In combination with the wealth of existing and upcoming spectro-photometric data in these fields, our IRAC mosaics will enable a wide range of galactic evolution and AGN studies. With that goal in mind, we make the combined IRAC mosaics and coverage maps of these three fields publicly available. counts show good agreement with previous studies.

astro-ph.GA↗

arXiv2023

BOLT: Fast Energy-based Controlled Text Generation with Tunable Biases

Energy-based models (EBMs) have gained popularity for controlled text generation due to their high applicability to a wide range of constraints. However, sampling from EBMs is non-trivial, as it often requires a large number of iterations to converge to plausible text, which slows down the decoding process and makes it less practical for real-world applications. In this work, we propose BOLT, which relies on tunable biases to directly adjust the language model's output logits. Unlike prior work, BOLT maintains the generator's autoregressive nature to assert a strong control on token-wise conditional dependencies and overall fluency, and thus converges faster. When compared with state-of-the-arts on controlled generation tasks using both soft constraints (e.g., sentiment control) and hard constraints (e.g., keyword-guided topic control), BOLT demonstrates significantly improved efficiency and fluency. On sentiment control, BOLT is 7x faster than competitive baselines, and more fluent in 74.4% of the evaluation samples according to human judges.

cs.CL↗

arXiv2023

Scalable and Sample Efficient Distributed Policy Gradient Algorithms in Multi-Agent Networked Systems

This paper studies a class of multi-agent reinforcement learning (MARL) problems where the reward that an agent receives depends on the states of other agents, but the next state only depends on the agent's own current state and action. We name it REC-MARL standing for REward-Coupled Multi-Agent Reinforcement Learning. REC-MARL has a range of important applications such as real-time access control and distributed power control in wireless networks. This paper presents a distributed policy gradient algorithm for REC-MARL. The proposed algorithm is distributed in two aspects: (i) the learned policy is a distributed policy that maps a local state of an agent to its local action and (ii) the learning/training is distributed, during which each agent updates its policy based on its own and neighbors' information. The learned algorithm achieves a stationary policy and its iterative complexity bounds depend on the dimension of local states and actions. The experimental results of our algorithm for the real-time access control and power control in wireless networks show that our policy significantly outperforms the state-of-the-art algorithms and well-known benchmarks.

cs.MA↗

arXiv2023

Rigorous justification of the hydrostatic approximation limit of viscous compressible flows

This paper considers the asymptotic limit of small aspect ratio between vertical and horizontal spatial scales for viscous isothermal compressible flows. In particular, it is observed that fast vertical acoustic waves arise and induce an averaging mechanism of the density in the vertical variable, which at the limit leads to the hydrostatic approximation of compressible flows, i.e., the compressible primitive equations of atmospheric dynamics. We justify the hydrostatic approximation for general as well as ``well-prepared'' initial data. The initial data is called well-prepared when it is close to the hydrostatic balance in a strong topology. Moreover, the convergence rate is calculated in the well-prepared initial data case in terms of the aspect ratio, as the latter goes to zero.

math.AP↗

arXiv2023

FolkScope: Intention Knowledge Graph Construction for E-commerce Commonsense Discovery

Understanding users' intentions in e-commerce platforms requires commonsense knowledge. In this paper, we present FolkScope, an intention knowledge graph construction framework to reveal the structure of humans' minds about purchasing items. As commonsense knowledge is usually ineffable and not expressed explicitly, it is challenging to perform information extraction. Thus, we propose a new approach that leverages the generation power of large language models~(LLMs) and human-in-the-loop annotation to semi-automatically construct the knowledge graph. LLMs first generate intention assertions via e-commerce-specific prompts to explain shopping behaviors, where the intention can be an open reason or a predicate falling into one of 18 categories aligning with ConceptNet, e.g., IsA, MadeOf, UsedFor, etc. Then we annotate plausibility and typicality labels of sampled intentions as training data in order to populate human judgments to all automatic generations. Last, to structurize the assertions, we propose pattern mining and conceptualization to form more condensed and abstract knowledge. Extensive evaluations and studies demonstrate that our constructed knowledge graph can well model e-commerce knowledge and have many potential applications.

cs.CL↗

arXiv2023

DiscoPrompt: Path Prediction Prompt Tuning for Implicit Discourse Relation Recognition

Implicit Discourse Relation Recognition (IDRR) is a sophisticated and challenging task to recognize the discourse relations between the arguments with the absence of discourse connectives. The sense labels for each discourse relation follow a hierarchical classification scheme in the annotation process (Prasad et al., 2008), forming a hierarchy structure. Most existing works do not well incorporate the hierarchy structure but focus on the syntax features and the prior knowledge of connectives in the manner of pure text classification. We argue that it is more effective to predict the paths inside the hierarchical tree (e.g., "Comparison -> Contrast -> however") rather than flat labels (e.g., Contrast) or connectives (e.g., however). We propose a prompt-based path prediction method to utilize the interactive information and intrinsic senses among the hierarchy in IDRR. This is the first work that injects such structure information into pre-trained language models via prompt tuning, and the performance of our solution shows significant and consistent improvement against competitive baselines.

cs.CL↗

arXiv2023

Walk4Me: Telehealth Community Mobility Assessment, An Automated System for Early Diagnosis and Disease Progression

We introduce Walk4Me, a telehealth community mobility assessment system designed to facilitate early diagnosis, severity, and progression identification. Our system achieves this by 1) enabling early diagnosis, 2) identifying early indicators of clinical severity, and 3) quantifying and tracking the progression of the disease across the ambulatory phase of the disease. To accomplish this, we employ an Artificial Intelligence (AI)-based detection of gait characteristics in patients and typically developing peers. Our system remotely and in real-time collects data from device sensors (e.g., acceleration from a mobile device, etc.) using our novel Walk4Me API. Our web application extracts temporal/spatial gait characteristics and raw data signal characteristics and then employs traditional machine learning and deep learning techniques to identify patterns that can 1) identify patients with gait disturbances associated with disease, 2) describe the degree of mobility limitation, and 3) identify characteristics that change over time with disease progression. We have identified several machine learning techniques that differentiate between patients and typically-developing subjects with 100% accuracy across the age range studied, and we have also identified corresponding temporal/spatial gait characteristics associated with each group. Our work demonstrates the potential of utilizing the latest advances in mobile device and machine learning technology to measure clinical outcomes regardless of the point of care, inform early clinical diagnosis and treatment decision-making, and monitor disease progression.

eess.SP↗

arXiv2023

MMPD: Multi-Domain Mobile Video Physiology Dataset

Remote photoplethysmography (rPPG) is an attractive method for noninvasive, convenient and concomitant measurement of physiological vital signals. Public benchmark datasets have served a valuable role in the development of this technology and improvements in accuracy over recent years.However, there remain gaps in the public datasets.First, despite the ubiquity of cameras on mobile devices, there are few datasets recorded specifically with mobile phone cameras. Second, most datasets are relatively small and therefore are limited in diversity, both in appearance (e.g., skin tone), behaviors (e.g., motion) and environment (e.g., lighting conditions). In an effort to help the field advance, we present the Multi-domain Mobile Video Physiology Dataset (MMPD), comprising 11 hours of recordings from mobile phones of 33 subjects. The dataset is designed to capture videos with greater representation across skin tone, body motion, and lighting conditions. MMPD is comprehensive with eight descriptive labels and can be used in conjunction with the rPPG-toolbox. The reliability of the dataset is verified by mainstream unsupervised methods and neural methods. The GitHub repository of our dataset: https://github.com/THU-CS-PI/MMPD_rPPG_dataset.

cs.CV↗

arXiv2023

On the Discredibility of Membership Inference Attacks

With the wide-spread application of machine learning models, it has become critical to study the potential data leakage of models trained on sensitive data. Recently, various membership inference (MI) attacks are proposed to determine if a sample was part of the training set or not. The question is whether these attacks can be reliably used in practice. We show that MI models frequently misclassify neighboring nonmember samples of a member sample as members. In other words, they have a high false positive rate on the subpopulations of the exact member samples that they can identify. We then showcase a practical application of MI attacks where this issue has a real-world repercussion. Here, MI attacks are used by an external auditor (investigator) to show to a judge/jury that an auditee unlawfully used sensitive data. Due to the high false positive rate of MI attacks on member's subpopulations, auditee challenges the credibility of the auditor by revealing the performance of the MI attacks on these subpopulations. We argue that current membership inference attacks can identify memorized subpopulations, but they cannot reliably identify which exact sample in the subpopulation was used during the training.

cs.CR↗