arXiv subjects

Kai Li

Publications and source records attributed to Kai Li.

At least 307 records · Page 17Linked to original sources

arXiv2023

Learning the black hole metric from holographic conductivity

We construct a neural network to learn the RN-AdS black hole metric based on the data of optical conductivity by holography. The linear perturbative equation for the Maxwell field is rewritten in terms of the optical conductivity such that the neural network is constructed based on the discretization of this differential equation. In contrast to all previous models in AdS/DL (deep learning) duality, the derivative of the metric function appears in the equation of motion and we propose distinct finite difference methods to discretize this function. The notion of the reduced conductivity is also proposed to avoid the divergence of the optical conductivity near the horizon.The dependence of the training outcomes on the location of the cutoff, the temperature as well as the frequency range is investigated in detail. This work provides a concrete example for the reconstruction of the bulk geometry with the given data on the boundary by deep learning.

hep-th↗

arXiv2023

Conditional Image-to-Video Generation with Latent Flow Diffusion Models

Conditional image-to-video (cI2V) generation aims to synthesize a new plausible video starting from an image (e.g., a person's face) and a condition (e.g., an action class label like smile). The key challenge of the cI2V task lies in the simultaneous generation of realistic spatial appearance and temporal dynamics corresponding to the given image and condition. In this paper, we propose an approach for cI2V using novel latent flow diffusion models (LFDM) that synthesize an optical flow sequence in the latent space based on the given condition to warp the given image. Compared to previous direct-synthesis-based works, our proposed LFDM can better synthesize spatial details and temporal motion by fully utilizing the spatial content of the given image and warping it in the latent space according to the generated temporally-coherent flow. The training of LFDM consists of two separate stages: (1) an unsupervised learning stage to train a latent flow auto-encoder for spatial content generation, including a flow predictor to estimate latent flow between pairs of video frames, and (2) a conditional learning stage to train a 3D-UNet-based diffusion model (DM) for temporal latent flow generation. Unlike previous DMs operating in pixel space or latent feature space that couples spatial and temporal information, the DM in our LFDM only needs to learn a low-dimensional latent flow space for motion generation, thus being more computationally efficient. We conduct comprehensive experiments on multiple datasets, where LFDM consistently outperforms prior arts. Furthermore, we show that LFDM can be easily adapted to new domains by simply finetuning the image decoder. Our code is available at https://github.com/nihaomiao/CVPR23_LFDM.

cs.CV↗

arXiv2023

On the Design and Training Strategies for RNN-based Online Neural Speech Separation Systems

While the performance of offline neural speech separation systems has been greatly advanced by the recent development of novel neural network architectures, there is typically an inevitable performance gap between the systems and their online variants. In this paper, we investigate how RNN-based offline neural speech separation systems can be changed into their online counterparts while mitigating the performance degradation. We decompose or reorganize the forward and backward RNN layers in a bidirectional RNN layer to form an online path and an offline path, which enables the model to perform both online and offline processing with a same set of model parameters. We further introduce two training strategies for improving the online model via either a pretrained offline model or a multitask training objective. Experiment results show that compared to the online models that are trained from scratch, the proposed layer decomposition and reorganization schemes and training strategies can effectively mitigate the performance gap between two RNN-based offline separation models and their online variants.

cs.SD↗

arXiv2023

Rethinking Lightweight Convolutional Neural Networks for Efficient and High-quality Pavement Crack Detection

Pixel-level road crack detection has always been a challenging task in intelligent transportation systems. Due to the external environments, such as weather, light, and other factors, pavement cracks often present low contrast, poor continuity, and different sizes in length and width. However, most of the existing studies pay less attention to crack data under different situations. Meanwhile, recent algorithms based on deep convolutional neural networks (DCNNs) have promoted the development of cutting-edge models for crack detection. Nevertheless, they usually focus on complex models for good performance, but ignore detection efficiency in practical applications. In this article, to address the first issue, we collected two new databases (i.e. Rain365 and Sun520) captured in rainy and sunny days respectively, which enrich the data of the open source community. For the second issue, we reconsider how to improve detection efficiency with excellent performance, and then propose our lightweight encoder-decoder architecture termed CarNet. Specifically, we introduce a novel olive-shaped structure for the encoder network, a light-weight multi-scale block and a new up-sampling method in the decoder network. Numerous experiments show that our model can better balance detection performance and efficiency compared with previous models. Especially, on the Sun520 dataset, our CarNet significantly advances the state-of-the-art performance with ODS F-score from 0.488 to 0.514. Meanwhile, it does so with an improved detection speed (104 frame per second) which is orders of magnitude faster than some recent DCNNs-based algorithms specially designed for crack detection.

cs.CV↗

arXiv2023

A Learned Cache Eviction Framework with Minimal Overhead

Recent work shows the effectiveness of Machine Learning (ML) to reduce cache miss ratios by making better eviction decisions than heuristics. However, state-of-the-art ML caches require many predictions to make an eviction decision, making them impractical for high-throughput caching systems. This paper introduces Machine learning At the Tail (MAT), a framework to build efficient ML-based caching systems by integrating an ML module with a traditional cache system based on a heuristic algorithm. MAT treats the heuristic algorithm as a filter to receive high-quality samples to train an ML model and likely candidate objects for evictions. We evaluate MAT on 8 production workloads, spanning storage, in-memory caching, and CDNs. The simulation experiments show MAT reduces the number of costly ML predictions-per-eviction from 63 to 2, while achieving comparable miss ratios to the state-of-the-art ML cache system. We compare a MAT prototype system with an LRU-based caching system in the same setting and show that they achieve similar request rates.

cs.OS↗

arXiv2023

Learning Trustworthy Model from Noisy Labels based on Rough Set for Surface Defect Detection

In the surface defect detection, there are some suspicious regions that cannot be uniquely classified as abnormal or normal. The annotating of suspicious regions is easily affected by factors such as workers' emotional fluctuations and judgment standard, resulting in noisy labels, which in turn leads to missing and false detections, and ultimately leads to inconsistent judgments of product quality. Unlike the usual noisy labels, the ones used for surface defect detection appear to be inconsistent rather than mislabeled. The noise occurs in almost every label and is difficult to correct or evaluate. In this paper, we proposed a framework that learns trustworthy models from noisy labels for surface defect defection. At first, to avoid the negative impact of noisy labels on the model, we represent the suspicious regions with consistent and precise elements at the pixel-level and redesign the loss function. Secondly, without changing network structure and adding any extra labels, pluggable spatially correlated Bayesian module is proposed. Finally, the defect discrimination confidence is proposed to measure the uncertainty, with which anomalies can be identified as defects. Our results indicate not only the effectiveness of the proposed method in learning from noisy labels, but also robustness and real-time performance.

cs.CV↗

arXiv2023

Adversarial Alignment for Source Free Object Detection

Source-free object detection (SFOD) aims to transfer a detector pre-trained on a label-rich source domain to an unlabeled target domain without seeing source data. While most existing SFOD methods generate pseudo labels via a source-pretrained model to guide training, these pseudo labels usually contain high noises due to heavy domain discrepancy. In order to obtain better pseudo supervisions, we divide the target domain into source-similar and source-dissimilar parts and align them in the feature space by adversarial learning. Specifically, we design a detection variance-based criterion to divide the target domain. This criterion is motivated by a finding that larger detection variances denote higher recall and larger similarity to the source domain. Then we incorporate an adversarial module into a mean teacher framework to drive the feature spaces of these two subsets indistinguishable. Extensive experiments on multiple cross-domain object detection datasets demonstrate that our proposed method consistently outperforms the compared SFOD methods.

cs.CV↗

arXiv2023

The study of eleven contact binaries with mass ratios less than 0.1

Multi-band photometric observations of eleven totally eclipsing contact binaries were carried out. Applying the Wilson-Devinney program, photometric solutions were obtained. There are two W-subtype systems, which are CRTS J133031.1+161202 and CRTS J154254.0+324652, and the rest systems are A-subtype systems. CRTS J154254.0+324652 has the highest fill-out factor with 94.3$\%$, and the lowest object is CRTS J155009.2+493639 with only 18.9$\%$. The mass ratios of the eleven systems are all less than 0.1, which means that they are extremely low mass ratio binary systems. We performed period variation investigation and found that the orbital periods of three systems decrease slowly, which may be caused by the angular momentum loss, and of six systems increase slowly, which indicates that the materials may transfer from the secondary component to the primary component. LAMOST low$-$resolution spectra of four objects were analyzed, and using the spectral subtraction technique, H$α$ emission line was detected, which means that the four objects exhibit chromospheric activity. In order to understand their evolutionary status, the mass-luminosity and mass-radius diagrams were plotted. The two diagrams indicate that the primary component is in the main sequence evolution stage, and the secondary component is above TAMS, indicating that they are over-luminous. To determine whether the eleven systems are in stable state, the ratio of spin angular momentum to orbital angular momentum ($J_{s}/J_{o}$) and the instability parameters were calculated, and we argued that CRTS J234634.7+222824 is on the verge of a merger.

astro-ph.SR↗

arXiv2022

Efficiently Hardening SGX Enclaves against Memory Access Pattern Attacks via Dynamic Program Partitioning

Intel SGX is known to be vulnerable to a class of practical attacks exploiting memory access pattern side-channels, notably page-fault attacks and cache timing attacks. A promising hardening scheme is to wrap applications in hardware transactions, enabled by Intel TSX, that return control to the software upon unexpected cache misses and interruptions so that the existing side-channel attacks exploiting these micro-architectural events can be detected and mitigated. However, existing hardening schemes scale only to small-data computation, with a typical working set smaller than one or few times (e.g., $8$ times) of a CPU data cache. This work tackles the data scalability and performance efficiency of security hardening schemes of Intel SGX enclaves against memory-access pattern side channels. The key insight is that the size of TSX transactions in the target computation is critical, both performance- and security-wise. Unlike the existing designs, this work dynamically partitions target computations to enlarge transactions while avoiding aborts, leading to lower performance overhead and improved side-channel security. We materialize the dynamic partitioning scheme and build a C++ library to monitor and model cache utilization at runtime. We further build a data analytical system using the library and implement various external oblivious algorithms. Performance evaluation shows that our work can effectively increase transaction size and reduce the execution time by up to two orders of magnitude compared with the state-of-the-art solutions.

cs.CR↗

arXiv2022

SSDNeRF: Semantic Soft Decomposition of Neural Radiance Fields

Neural Radiance Fields (NeRFs) encode the radiance in a scene parameterized by the scene's plenoptic function. This is achieved by using an MLP together with a mapping to a higher-dimensional space, and has been proven to capture scenes with a great level of detail. Naturally, the same parameterization can be used to encode additional properties of the scene, beyond just its radiance. A particularly interesting property in this regard is the semantic decomposition of the scene. We introduce a novel technique for semantic soft decomposition of neural radiance fields (named SSDNeRF) which jointly encodes semantic signals in combination with radiance signals of a scene. Our approach provides a soft decomposition of the scene into semantic parts, enabling us to correctly encode multiple semantic classes blending along the same direction -- an impossible feat for existing methods. Not only does this lead to a detailed, 3D semantic representation of the scene, but we also show that the regularizing effects of the MLP used for encoding help to improve the semantic representation. We show state-of-the-art segmentation and reconstruction results on a dataset of common objects and demonstrate how the proposed approach can be applied for high quality temporally consistent video editing and re-compositing on a dataset of casually captured selfie videos.

cs.CV↗

arXiv2022

Machine Learning-Aided Operations and Communications of Unmanned Aerial Vehicles: A Contemporary Survey

The ongoing amalgamation of UAV and ML techniques is creating a significant synergy and empowering UAVs with unprecedented intelligence and autonomy. This survey aims to provide a timely and comprehensive overview of ML techniques used in UAV operations and communications and identify the potential growth areas and research gaps. We emphasise the four key components of UAV operations and communications to which ML can significantly contribute, namely, perception and feature extraction, feature interpretation and regeneration, trajectory and mission planning, and aerodynamic control and operation. We classify the latest popular ML tools based on their applications to the four components and conduct gap analyses. This survey also takes a step forward by pointing out significant challenges in the upcoming realm of ML-aided automated UAV operations and communications. It is revealed that different ML techniques dominate the applications to the four key modules of UAV operations and communications. While there is an increasing trend of cross-module designs, little effort has been devoted to an end-to-end ML framework, from perception and feature extraction to aerodynamic control and operation. It is also unveiled that the reliability and trust of ML in UAV operations and applications require significant attention before full automation of UAVs and potential cooperation between UAVs and humans come to fruition.

cs.RO↗

arXiv2022

Recovering Private Text in Federated Learning of Language Models

Federated learning allows distributed users to collaboratively train a model while keeping each user's data private. Recently, a growing body of work has demonstrated that an eavesdropping attacker can effectively recover image data from gradients transmitted during federated learning. However, little progress has been made in recovering text data. In this paper, we present a novel attack method FILM for federated learning of language models (LMs). For the first time, we show the feasibility of recovering text from large batch sizes of up to 128 sentences. Unlike image-recovery methods that are optimized to match gradients, we take a distinct approach that first identifies a set of words from gradients and then directly reconstructs sentences based on beam search and a prior-based reordering strategy. We conduct the FILM attack on several large-scale datasets and show that it can successfully reconstruct single sentences with high fidelity for large batch sizes and even multiple sentences if applied iteratively. We evaluate three defense methods: gradient pruning, DPSGD, and a simple approach to freeze word embeddings that we propose. We show that both gradient pruning and DPSGD lead to a significant drop in utility. However, if we fine-tune a public pre-trained LM on private text without updating word embeddings, it can effectively defend the attack with minimal data utility loss. Together, we hope that our results can encourage the community to rethink the privacy concerns of LM training and its standard practices in the future.

cs.CL↗

arXiv2022

Photometric and spectroscopic studies of the long period low mass ratio deep contact binary KN Per

The photometric analysis and spectroscopic study of the long period low mass ratio deep contact binary KN Per were executed. The light curves of BV(RI)$_c$-band were from the Ningbo Bureau of Education and Xinjiang Observatory Telescope (NEXT) at the Xingming Observatory. Through the analysis of Wilson-Devinney (W-D) program, KN Per was found as an A-type low mass ratio deep contact binary (q=0.236, f=53.4\%). A cool spot applied on the primary component was introduced to explain the unequal maxima of the light curve. Based on the O-C analysis, we found that the rate of the increasing orbital period is $\dot{P}$ = 5.12 $\pm$ (0.30) $\times$ 10$^{-7}$ d/yr, meaning the mass transfer from the secondary component to the primary one. By analyzing the spectroscopic data, we detected chromospheric activity emission line indicators, which is corresponding to the light-curve analysis. 71 long period (P $>$ 0.5 days) contact binaries including our target were collected. The evolutionary states of all collected stars were investigated by the illustrations of mass-radius, mass-luminosity, and log M$_{T}$ - log J$_{o}$. The relations of some physical parameters were also determined. With the instability parameters of KN Per, we determined that it is a stable contact binary system at present.

astro-ph.SR↗

arXiv2022

Deep Spectro-temporal Artifacts for Detecting Synthesized Speech

The Audio Deep Synthesis Detection (ADD) Challenge has been held to detect generated human-like speech. With our submitted system, this paper provides an overall assessment of track 1 (Low-quality Fake Audio Detection) and track 2 (Partially Fake Audio Detection). In this paper, spectro-temporal artifacts were detected using raw temporal signals, spectral features, as well as deep embedding features. To address track 1, low-quality data augmentation, domain adaptation via finetuning, and various complementary feature information fusion were aggregated in our system. Furthermore, we analyzed the clustering characteristics of subsystems with different features by visualization method and explained the effectiveness of our proposed greedy fusion strategy. As for track 2, frame transition and smoothing were detected using self-supervised learning structure to capture the manipulation of PF attacks in the time domain. We ranked 4th and 5th in track 1 and track 2, respectively.

cs.SD↗

arXiv2022

The First Photometric and Orbital Period Investigation of an Extremely Low Mass Ratio Contact Binary with a Sudden Period Change, TYC 4002-2628-1

Photometric observations for the totally eclipsing binary system TYC 4002-2628-1, were obtained between November 2020 and November 2021. To determine the stellar atmospheric parameters, a spectral image was taken with the 2.16 m telescope at National Astronomical Observatory of China (NAOC). TYC 4002-2628-1 is a low-amplitude (about 0.15 mag for $V$ band), short-period (0.3670495 d), contact eclipsing binary with a total secondary eclipse. Intrinsic light curve variations and the reversal of the O'Connell effect are detected in the light curves, which may be due to spot activity. Based on the photometric solutions derived from the multi-band time series light curves, TYC 4002-2628-1 is an extremely low mass ratio contact binary with a mass ratio of $q\sim$ 0.0482 and a fill-out factor of $f\sim5\%$. By analyzing the $O-C$ variations, we find that its orbital period remains unchanged when BJD < 2458321 . Then the orbital period changed suddenly around BJD 2458743 and has an increasing rate of $dP/dt=1.62\times{10^{-5}}day\cdot yr^{-1}=140$ $second\cdot century^{-1}$ . If confirmed, TYC 4002-2628-1 would be the contact binary with the highest orbital period increasing rate so far. By investigating the ratio of orbital angular momentum to the spin angular momentum ( $J_{orb}$/$J_{spin}$ $<3$) , the instability mass ratio ($q_{inst}/q=1.84$) and the instability separation ($A_{inst}/A=1.35$), TYC 4002-2628-1 can be regarded as a merger candidate.

astro-ph.SR↗

arXiv2022

A Review on Method Entities in the Academic Literature: Extraction, Evaluation, and Application

In scientific research, the method is an indispensable means to solve scientific problems and a critical research object. With the advancement of sciences, many scientific methods are being proposed, modified, and used in academic literature. The authors describe details of the method in the abstract and body text, and key entities in academic literature reflecting names of the method are called method entities. Exploring diverse method entities in a tremendous amount of academic literature helps scholars understand existing methods, select the appropriate method for research tasks, and propose new methods. Furthermore, the evolution of method entities can reveal the development of a discipline and facilitate knowledge discovery. Therefore, this article offers a systematic review of methodological and empirical works focusing on extracting method entities from full-text academic literature and efforts to build knowledge services using these extracted method entities. Definitions of key concepts involved in this review were first proposed. Based on these definitions, we systematically reviewed the approaches and indicators to extract and evaluate method entities, with a strong focus on the pros and cons of each approach. We also surveyed how extracted method entities are used to build new applications. Finally, limitations in existing works as well as potential next steps were discussed.

cs.DL↗

arXiv2022

Topological Quantum Phase Transitions in Metallic Shiba Lattices

Shiba bands formed by overlapping Yu-Shiba-Rusinov subgap states in magnetic impurities on a superconductor play an important role in topological superconductors. Here, we theoretically demonstrate the existence of a new type of Shiba bands (dubbed topological Shiba metal) on a magnetically doped $s$-wave superconducting surface with Rashba spin-orbit coupling in the presence of a weak in-plane magnetic field. Such topological gapless Shiba bands develop from gapped Shiba bands through Lifshitz phase transitions accompanied by second-order quantum phase transitions for the intrinsic thermal Hall conductance. We also find a mechanism in Shiba lattices that protects the first-order quantum phase transitions for the intrinsic thermal Hall conductance. Due to the long-range hopping in Shiba lattices, the topological Shiba metal exhibits intrinsic thermal Hall conductance with large nonquantized values. As a consequence, there emerge a large number of second-order quantum phase transitions.

cond-mat.mes-hall↗

arXiv2022

Extremely low mass ratio contact binaries -- I. the first photometric and spectroscopic investigations of ten systems

The photometric and spectroscopic investigations of ten contact binaries were presented for the first time. It is discovered that the mass ratios of all the ten targets are smaller than 0.15, they are extremely low mass ratio contact binaries. Seven of them are deep contact binaries, two are medium contact binaries, while only one is a shallow contact system. Five of them show O'Connell effect, and a dark spot on one of the two components can lead to a good fit of the asymmetric light curves. The orbital period studies of the ten binaries reveal that they all exhibit long-term period changes, six of them are increasing, while the others are shrinking. The LAMOST spectra were analyzed by the spectral subtraction method, and all the ten targets exhibit excess emissions in the H$_α$ line, indicating chromospheric activity. The evolutionary states of the two components of the ten binaries were studied, and it is found that their evolutionary states are identical to those of the other contact binaries. Based on the study of the relation between orbital angular momentum and total mass, we discovered the ten systems may be at the late evolutionary stage of a contact binary. The initial masses of the two components and the ages of them were obtained. By calculating the instability parameters, we found that the ten contact binaries are relatively stable at present.

astro-ph.SR↗