arXiv subjects

Yao Li

Publications and source records attributed to Yao Li.

At least 127 records · Page 7Linked to original sources

arXiv2024

Adversarial Examples Detection with Bayesian Neural Network

In this paper, we propose a new framework to detect adversarial examples motivated by the observations that random components can improve the smoothness of predictors and make it easier to simulate the output distribution of a deep neural network. With these observations, we propose a novel Bayesian adversarial example detector, short for BATer, to improve the performance of adversarial example detection. Specifically, we study the distributional difference of hidden layer output between natural and adversarial examples, and propose to use the randomness of the Bayesian neural network to simulate hidden layer output distribution and leverage the distribution dispersion to detect adversarial examples. The advantage of a Bayesian neural network is that the output is stochastic while a deep neural network without random components does not have such characteristics. Empirical results on several benchmark datasets against popular attacks show that the proposed BATer outperforms the state-of-the-art detectors in adversarial example detection.

stat.ML↗

arXiv2024

DeepSeek LLM: Scaling Open-Source Language Models with Longtermism

The rapid development of open-source large language models (LLMs) has been truly remarkable. However, the scaling law described in previous literature presents varying conclusions, which casts a dark cloud over scaling LLMs. We delve into the study of scaling laws and present our distinctive findings that facilitate scaling of large scale models in two commonly used open-source configurations, 7B and 67B. Guided by the scaling laws, we introduce DeepSeek LLM, a project dedicated to advancing open-source language models with a long-term perspective. To support the pre-training phase, we have developed a dataset that currently consists of 2 trillion tokens and is continuously expanding. We further conduct supervised fine-tuning (SFT) and Direct Preference Optimization (DPO) on DeepSeek LLM Base models, resulting in the creation of DeepSeek Chat models. Our evaluation results demonstrate that DeepSeek LLM 67B surpasses LLaMA-2 70B on various benchmarks, particularly in the domains of code, mathematics, and reasoning. Furthermore, open-ended evaluations reveal that DeepSeek LLM 67B Chat exhibits superior performance compared to GPT-3.5.

cs.CL↗

arXiv2023

Nanospheres with Patches Arranged in Polyhedrons from Self-Assembly of Solution-State Diblock Copolymers under Spherical Confinement

Self-assembly of sphere-forming solution-state amphiphilic diblock copolymers under spherical nanopore confinement is investigated using a simulated annealing technique. For two types of cases of different pore-surface/copolymer interactions, sequences of self-assembled patchy nanospheres are obtained, and phase diagrams are constructed. Self-assembled patchy nanospheres with 1-21 solvophobic domains are observed. The outermost solvophobic domains (patches) are packed into various polyhedrons when their number is larger than 3, where three Platonic solids of a regular tetrahedron, an octahedron, and an icosahedron and seven Johnson solids of J12, J13, J17, J50, J51, J86, and J87 are identified. In addition, another Johnson solid of J84 is identified in a structure with two categories of B-domains. These polyhedrons have all or most of their faces in a triangular shape, and hence, they are closer to spherical in shape, which may relieve the chain stretching. Nanospheres with 1, 4, 6, 9, and 12 numbers of patches occur in relatively large windows in the phase diagrams of both types of cases. In one of the two types of cases, all nanospheres with any number of 1-14 patches occur in the phase diagram, whereas in the other type of cases, nanospheres with 2, 3, 5, 11, and 13 numbers of patches are absent in the phase diagram. Furthermore, at a given pore size, the number of patches changes nonmonotonically or is unchanged with an increase in the strength of the pore-surface/copolymer interactions for one type or the other type of case, respectively. Quantitative calculations are performed to elucidate mechanisms of the window size in the phase diagrams of nanospheres with different numbers of patches and structure details.

cond-mat.soft↗

arXiv2023

Self-assembly of Colloids with Competing Interactions Confined in Spheres

At low temperatures, colloidal particles with short-range attractive and long-range repulsive interactions can form various periodic microphases in bulk.In this paper, we investigate the self-assembly behaviour of colloids with competing interactions under spherical confinement by conducting molecular dynamics simulations. We find that the cluster, mixture, cylindrical, perforated lamellar and lamellar structures can be obtained, but the details of the ordered structures are different from those in bulk systems. Interestingly, the system tends to form more perforated structures when confined in smaller spheres. The mechanism behind this phenomenon is the relationship between the energy of the ordered structures and the bending of the confinement wall, which is different from the mechanism in copolymer systems.

cond-mat.soft↗

arXiv2023

Older and younger adults are influenced differently by dark pattern designs

Considering that prior research has found older users undergo a different privacy decision-making process compared to younger adults, more research is needed to inform the behavioral privacy disclosure effects of these strategies for different age groups. To address this gap, we used an existing dataset of an experiment with a photo-tagging Facebook application. This experiment had a 2x2x5 between-subjects design where the manipulations were common dark pattern design strategies: framing (positive vs. negative), privacy defaults (opt-in vs. opt-out), and justification messages (positive normative, negative normative, positive rationale, negative rationale, none). We compared older (above 65 years old, N=44) and young adults (18 to 25 years old, N=162) privacy concerns and disclosure behaviors (i.e., accepting or refusing automated photo tagging) in the scope of dark pattern design. Overall, we find support for the effectiveness of dark pattern designs in the sense that positive framing and opt-out privacy defaults significantly increased disclosure behavior, while negative justification messages significantly decreased privacy concerns. Regarding older adults, our results show that certain dark patterns do lead to more disclosure than for younger adults, but also to increased privacy concerns for older adults than for younger.

cs.HC↗

arXiv2023

Manipulating polariton condensates by Rashba-Dresselhaus coupling at room temperature

The spin-orbit coupling plays an important role in the spin Hall effect and the topological insulators. In addition, the spin-orbit coupled Bose-Einstein condensates show remarkable quantum many-body phase transition. In this work we tune the exciton polariton condensate by virtue of the Rashba-Dresselhaus (RD) spin-orbit coupling in a liquid-crystal filled microcavity where perovskite CsPbBr3 microplates act as the gain material at room temperature. We realize an artificial gauge field on the CsPbBr3 exciton polariton condensate, which splits the condensates with opposite spins in both momentum and real spaces. Our work paves the way to manipulate the exciton polariton condensate with a synthetic gauge field based on the RD spin-orbit coupling at room temperature.

cond-mat.quant-gas↗

arXiv2023

MiChao-HuaFen 1.0: A Specialized Pre-trained Corpus Dataset for Domain-specific Large Models

With the advancement of deep learning technologies, general-purpose large models such as GPT-4 have demonstrated exceptional capabilities across various domains. Nevertheless, there remains a demand for high-quality, domain-specific outputs in areas like healthcare, law, and finance. This paper first evaluates the existing large models for specialized domains and discusses their limitations. To cater to the specific needs of certain domains, we introduce the ``MiChao-HuaFen 1.0'' pre-trained corpus dataset, tailored for the news and governmental sectors. The dataset, sourced from publicly available internet data from 2022, underwent multiple rounds of cleansing and processing to ensure high quality and reliable origins, with provisions for consistent and stable updates. This dataset not only supports the pre-training of large models for Chinese vertical domains but also aids in propelling deep learning research and applications in related fields.

cs.CL↗

arXiv2023

A Tale of Two Cultures: Comparing Interpersonal Information Disclosure Norms on Twitter

We present an exploration of cultural norms surrounding online disclosure of information about one's interpersonal relationships (such as information about family members, colleagues, friends, or lovers) on Twitter. The literature identifies the cultural dimension of individualism versus collectivism as being a major determinant of offline communication differences in terms of emotion, topic, and content disclosed. We decided to study whether such differences also occur online in context of Twitter when comparing tweets posted in an individualistic (U.S.) versus a collectivist (India) society. We collected more than 2 million tweets posted in the U.S. and India over a 3 month period which contain interpersonal relationship keywords. A card-sort study was used to develop this culturally-sensitive saturated taxonomy of keywords that represent interpersonal relationships (e.g., ma, mom, mother). Then we developed a high-accuracy interpersonal disclosure detector based on dependency-parsing (F1-score: 86%) to identify when the words refer to a personal relationship of the poster (e.g., "my mom" as opposed to "a mom"). This allowed us to identify the 400K+ tweets in our data set which actually disclose information about the poster's interpersonal relationships. We used a mixed methods approach to analyze these tweets (e.g., comparing the amount of joy expressed about one's family) and found differences in emotion, topic, and content disclosed between tweets from the U.S. versus India. Our analysis also reveals how a combination of qualitative and quantitative methods are needed to uncover these differences; Using just one or the other can be misleading. This study extends the prior literature on Multi-Party Privacy and provides guidance for researchers and designers of culturally-sensitive systems.

cs.HC↗

arXiv2023

Longitudinal momentum fraction of heavy-flavor mesons in jets in high-energy nuclear collisions

Heavy-flavor jets are powerful tools to gain insight into the in-medium partonic energy-loss mechanisms and the quark-gluon plasma's (QGP) transport properties in high-energy nuclear collisions. In this work we present the first theoretical study of the longitudinal momentum fraction $z_{||}$ carried by heavy-flavor mesons in jets in Pb+Pb collisions at $\sqrt{s_{\rm NN}}$ = 5.02 TeV. The p+p baseline is provided by POWHEG+PYTHIA8, which matches the next-to-leading-order hard processes with the parton shower. We employ a Monte Carlo transport model, which considers the collisional and radiative partonic energy loss, to simulate the evolution of heavy-flavor jets in the expanding QGP medium. We observe steeper $z_{||}$ distributions of $\rm{B^0}$ jets compared to those of $\rm{D^0}$ jets at the same kinematics region in p+p collisions, which may be a hint of the harder jet fragmentation function of b jets compared to c jets in vacuum. In A+A collisions, it is shown that the jet quenching effect would generally decrease the values of $z_{||}$. We have made a systematical study on how several factors, including jet $p_{\rm{T}}$, jet radius $R$, and collision centrality, would influence the medium modification of $z_{||}$ distributions of a $\rm{D^0}$ jet. In addition, we predict visibly stronger nuclear modifications of $\rm{B^0}$-jet $z_{||}$ distributions compared to a $\rm{D^0}$ jet within the same $p_{\rm T}$ windows as a result of the much steeper initial $z_{||}$ distribution of the $\rm{B^0}$ jet in vacuum.

hep-ph↗

arXiv2023

A Review of Adversarial Attacks in Computer Vision

Deep neural networks have been widely used in various downstream tasks, especially those safety-critical scenario such as autonomous driving, but deep networks are often threatened by adversarial samples. Such adversarial attacks can be invisible to human eyes, but can lead to DNN misclassification, and often exhibits transferability between deep learning and machine learning models and real-world achievability. Adversarial attacks can be divided into white-box attacks, for which the attacker knows the parameters and gradient of the model, and black-box attacks, for the latter, the attacker can only obtain the input and output of the model. In terms of the attacker's purpose, it can be divided into targeted attacks and non-targeted attacks, which means that the attacker wants the model to misclassify the original sample into the specified class, which is more practical, while the non-targeted attack just needs to make the model misclassify the sample. The black box setting is a scenario we will encounter in practice.

cs.CV↗

arXiv2023

DBFed: Debiasing Federated Learning Framework based on Domain-Independent

As digital transformation continues, enterprises are generating, managing, and storing vast amounts of data, while artificial intelligence technology is rapidly advancing. However, it brings challenges in information security and data security. Data security refers to the protection of digital information from unauthorized access, damage, theft, etc. throughout its entire life cycle. With the promulgation and implementation of data security laws and the emphasis on data security and data privacy by organizations and users, Privacy-preserving technology represented by federated learning has a wide range of application scenarios. Federated learning is a distributed machine learning computing framework that allows multiple subjects to train joint models without sharing data to protect data privacy and solve the problem of data islands. However, the data among multiple subjects are independent of each other, and the data differences in quality may cause fairness issues in federated learning modeling, such as data bias among multiple subjects, resulting in biased and discriminatory models. Therefore, we propose DBFed, a debiasing federated learning framework based on domain-independent, which mitigates model bias by explicitly encoding sensitive attributes during client-side training. This paper conducts experiments on three real datasets and uses five evaluation metrics of accuracy and fairness to quantify the effect of the model. Most metrics of DBFed exceed those of the other three comparative methods, fully demonstrating the debiasing effect of DBFed.

cs.LG↗

arXiv2023

Evaluating Similitude and Robustness of Deep Image Denoising Models via Adversarial Attack

Deep neural networks (DNNs) have shown superior performance comparing to traditional image denoising algorithms. However, DNNs are inevitably vulnerable while facing adversarial attacks. In this paper, we propose an adversarial attack method named denoising-PGD which can successfully attack all the current deep denoising models while keep the noise distribution almost unchanged. We surprisingly find that the current mainstream non-blind denoising models (DnCNN, FFDNet, ECNDNet, BRDNet), blind denoising models (DnCNN-B, Noise2Noise, RDDCNN-B, FAN), plug-and-play (DPIR, CurvPnP) and unfolding denoising models (DeamNet) almost share the same adversarial sample set on both grayscale and color images, respectively. Shared adversarial sample set indicates that all these models are similar in term of local behaviors at the neighborhood of all the test samples. Thus, we further propose an indicator to measure the local similarity of models, called robustness similitude. Non-blind denoising models are found to have high robustness similitude across each other, while hybrid-driven models are also found to have high robustness similitude with pure data-driven non-blind denoising models. According to our robustness assessment, data-driven non-blind denoising models are the most robust. We use adversarial training to complement the vulnerability to adversarial attacks. Moreover, the model-driven image denoising BM3D shows resistance on adversarial attacks.

cs.CV↗

arXiv2023

Artificial Neural Network Prediction of COVID-19 Daily Infection Count

It is well known that the confirmed COVID-19 infection is only a fraction of the true fraction. In this paper we use an artificial neural network to learn the connection between the confirmed infection count, the testing data, and the true infection count. The true infection count in the training set is obtained by backcasting from the death count and the infection fatality ratio (IFR). Multiple factors are taken into consideration in the estimation of IFR. We also calibrate the recovered true COVID-19 case count with an SEIR model.

q-bio.PE↗

arXiv2023

Bi-LRFusion: Bi-Directional LiDAR-Radar Fusion for 3D Dynamic Object Detection

LiDAR and Radar are two complementary sensing approaches in that LiDAR specializes in capturing an object's 3D shape while Radar provides longer detection ranges as well as velocity hints. Though seemingly natural, how to efficiently combine them for improved feature representation is still unclear. The main challenge arises from that Radar data are extremely sparse and lack height information. Therefore, directly integrating Radar features into LiDAR-centric detection networks is not optimal. In this work, we introduce a bi-directional LiDAR-Radar fusion framework, termed Bi-LRFusion, to tackle the challenges and improve 3D detection for dynamic objects. Technically, Bi-LRFusion involves two steps: first, it enriches Radar's local features by learning important details from the LiDAR branch to alleviate the problems caused by the absence of height information and extreme sparsity; second, it combines LiDAR features with the enhanced Radar features in a unified bird's-eye-view representation. We conduct extensive experiments on nuScenes and ORR datasets, and show that our Bi-LRFusion achieves state-of-the-art performance for detecting dynamic objects. Notably, Radar data in these two datasets have different formats, which demonstrates the generalizability of our method. Codes are available at https://github.com/JessieW0806/BiLRFusion.

cs.CV↗

arXiv2023

CTSN: Predicting Cloth Deformation for Skeleton-based Characters with a Two-stream Skinning Network

We present a novel learning method to predict the cloth deformation for skeleton-based characters with a two-stream network. The characters processed in our approach are not limited to humans, and can be other skeletal-based representations of non-human targets such as fish or pets. We use a novel network architecture which consists of skeleton-based and mesh-based residual networks to learn the coarse and wrinkle features as the overall residual from the template cloth mesh. Our network is used to predict the deformation for loose or tight-fitting clothing or dresses. We ensure that the memory footprint of our network is low, and thereby result in reduced storage and computational requirements. In practice, our prediction for a single cloth mesh for the skeleton-based character takes about 7 milliseconds on an NVIDIA GeForce RTX 3090 GPU. Compared with prior methods, our network can generate fine deformation results with details and wrinkles.

cs.GR↗

arXiv2023

Accelerating Dataset Distillation via Model Augmentation

Dataset Distillation (DD), a newly emerging field, aims at generating much smaller but efficient synthetic training datasets from large ones. Existing DD methods based on gradient matching achieve leading performance; however, they are extremely computationally intensive as they require continuously optimizing a dataset among thousands of randomly initialized models. In this paper, we assume that training the synthetic data with diverse models leads to better generalization performance. Thus we propose two model augmentation techniques, i.e. using early-stage models and parameter perturbation to learn an informative synthetic set with significantly reduced training cost. Extensive experiments demonstrate that our method achieves up to 20x speedup and comparable performance on par with state-of-the-art methods.

cs.LG↗

arXiv2023

You Need Multiple Exiting: Dynamic Early Exiting for Accelerating Unified Vision Language Model

Large-scale Transformer models bring significant improvements for various downstream vision language tasks with a unified architecture. The performance improvements come with increasing model size, resulting in slow inference speed and increased cost for severing. While some certain predictions benefit from the full complexity of the large-scale model, not all of inputs need the same amount of computation to conduct, potentially leading to computation resource waste. To handle this challenge, early exiting is proposed to adaptively allocate computational power in term of input complexity to improve inference efficiency. The existing early exiting strategies usually adopt output confidence based on intermediate layers as a proxy of input complexity to incur the decision of skipping following layers. However, such strategies cannot apply to encoder in the widely-used unified architecture with both encoder and decoder due to difficulty of output confidence estimation in the encoder. It is suboptimal in term of saving computation power to ignore the early exiting in encoder component. To handle this challenge, we propose a novel early exiting strategy for unified visual language models, which allows dynamically skip the layers in encoder and decoder simultaneously in term of input layer-wise similarities with multiple times of early exiting, namely \textbf{MuE}. By decomposing the image and text modalities in the encoder, MuE is flexible and can skip different layers in term of modalities, advancing the inference efficiency while minimizing performance drop. Experiments on the SNLI-VE and MS COCO datasets show that the proposed approach MuE can reduce expected inference time by up to 50\% and 40\% while maintaining 99\% and 96\% performance respectively.

cs.CV↗

arXiv2023

Collective-motion-enhanced acceleration sensing via an optically levitated microsphere array

Optically levitated microspheres are an excellent candidate for force and acceleration sensing. Here, we propose an acceleration sensing protocol based on an optically levitated microsphere array (MSA). The system consists of an $N$-microsphere array levitated in a driven optical cavity via holographic optical tweezers. By positioning the microspheres suitably relative to the cavity, only one of the collective modes of the MSA is coupled to the cavity mode. The optomechanical interaction encodes the information of acceleration acting on the MSA onto the intracavity photons, which can then be detected directly at the output of the cavity. The optically levitated MSA forms an effective large mass-distributed particle, which not only circumvents the problem of levitating a large mass microsphere but also results in a significant improvement of sensitivity. Compared with the traditional single-microsphere measurement scheme, our method presents an improvement in sensitivity by a factor of $\sqrt{N}$.

physics.optics↗