Search

We consider the problem of user-adaptive 3D gaze estimation. The performance of person-independent gaze estimation is limited due to interpersonal anatomical differences. Our goal is to provide a personalized gaze estimation model specifically adapted to a target user. Previous work on user-adaptive gaze estimation requires some labeled images of the target person data to fine-tune the model at test time. However, this can be unrealistic in real-world applications, since it is cumbersome for an end-user to provide labeled images. In addition, previous work requires the training data to have both gaze labels and person IDs. This data requirement makes it infeasible to use some of the available data. To tackle these challenges, this paper proposes a new problem called efficient label-free user adaptation in gaze estimation. Our model only needs a few unlabeled images of a target user for the model adaptation. During offline training, we have some labeled source data without person IDs and some unlabeled person-specific data. Our proposed method uses a meta-learning approach to learn how to adapt to a new user with only a few unlabeled images. Our key technical innovation is to use a generalization bound from domain adaptation to define the loss function in meta-learning, so that our method can effectively make use of both the labeled source data and the unlabeled person-specific data during training. Extensive experiments validate the effectiveness of our method on several challenging benchmarks., Comment: This paper has been accepted by IJCAI'24

Since Newton's time, deterministic causality has been considered a crucial prerequisite in any fundamental theory in physics. In contrast, the present work investigates stochastic dynamical models for motion in one spatial dimension, in which Newtonian mechanics becomes an emergent property: We present a coherent theory in which a Hamilton-Jacobi equation (HJE) emerges in a description of the evolution of entropy $-\phi(x,t)=\epsilon \log$(Probability) of a system under observation and in the limit of large information extent $\epsilon^{-1}$ in homogeneous space and time. The variable $\phi$ represents a non-random high-order statistical concept that is distinct from probability itself as $\epsilon=0$; the HJE embodies an emergent law of deterministic causality in continuous space and time with an Imaginary Scale symmetry $(t,x,\phi)\leftrightarrow (it,ix,i\phi)$. $\phi(x,t)$ exhibits a nonlinear wave phenomenon with a mathematical singularity in finite time, overcoming which we introduce viscosity $\epsilon(\partial^2\phi/\partial x^2)$ and wave $i\epsilon(\partial^2 \phi/\partial x^2)$ perturbations, articulating dissipation and conservation, which break the Imaginary Scale symmetry: They lead to the Brownian motion and Schr\"{o}dinger's equation of motion, respectively. Last but not least, Lagrange's action in classical mechanics acquires an entropic interpretation and Hamilton's principle is established., Comment: 10 pages

In this paper, we introduce a conservative Crank-Nicolson-type finite difference schemes for the regularized logarithmic Schr\"{o}dinger equation (RLSE) with Dirac delta potential in 1D. The regularized logarithmic Schr\"{o}dinger equation with a small regularized parameter $0<\eps \ll 1$ is adopted to approximate the logarithmic Schr\"{o}dinger equation (LSE) with linear convergence rate $O(\eps)$. The numerical method can be used to avoid numerical blow-up and/or to suppress round-off error due to the logarithmic nonlinearity in LSE. Then, by using domain-decomposition technique, we can transform the original problem into an interface problem. Different treatments on the interface conditions lead to different discrete schemes and it turns out that a simple discrete approximation of the Dirac potential coincides with one of the conservative finite difference schemes. The optimal $H^1$ error estimates and the conservative properties of the finite difference schemes are investigated. The Crank-Nicolson finite difference methods enjoy the second-order convergence rate in time and space. Numerical examples are provided to support our analysis and show the accuracy and efficiency of the numerical method., Comment: 20 pages, 3 figures

This paper reviews the NTIRE 2024 Challenge on Shortform UGC Video Quality Assessment (S-UGC VQA), where various excellent solutions are submitted and evaluated on the collected dataset KVQ from popular short-form video platform, i.e., Kuaishou/Kwai Platform. The KVQ database is divided into three parts, including 2926 videos for training, 420 videos for validation, and 854 videos for testing. The purpose is to build new benchmarks and advance the development of S-UGC VQA. The competition had 200 participants and 13 teams submitted valid solutions for the final testing phase. The proposed solutions achieved state-of-the-art performances for S-UGC VQA. The project can be found at https://github.com/lixinustc/KVQChallenge-CVPR-NTIRE2024., Comment: Accepted by CVPR2024 Workshop. The challenge report for CVPR NTIRE2024 Short-form UGC Video Quality Assessment Challenge

Deep learning has achieved remarkable progress in various applications, heightening the importance of safeguarding the intellectual property (IP) of well-trained models. It entails not only authorizing usage but also ensuring the deployment of models in authorized data domains, i.e., making models exclusive to certain target domains. Previous methods necessitate concurrent access to source training data and target unauthorized data when performing IP protection, making them risky and inefficient for decentralized private data. In this paper, we target a practical setting where only a well-trained source model is available and investigate how we can realize IP protection. To achieve this, we propose a novel MAsk Pruning (MAP) framework. MAP stems from an intuitive hypothesis, i.e., there are target-related parameters in a well-trained model, locating and pruning them is the key to IP protection. Technically, MAP freezes the source model and learns a target-specific binary mask to prevent unauthorized data usage while minimizing performance degradation on authorized data. Moreover, we introduce a new metric aimed at achieving a better balance between source and target performance degradation. To verify the effectiveness and versatility, we have evaluated MAP in a variety of scenarios, including vanilla source-available, practical source-free, and challenging data-free. Extensive experiments indicate that MAP yields new state-of-the-art performance., Comment: Accepted to CVPR 2024

Synchronization phenomena are pervasive in coupled nonlinear systems across the natural world and engineering domains. Understanding how to dynamically identify the parameter space (or network structure) of coupled nonlinear systems in a synchronized state is crucial for the study of system synchronization. To address the challenge of achieving stable synchronization in coupled nonlinear systems, we develop a set of mathematical optimization techniques for dynamic learning of synchronization (DLS) inspired by machine learning. This technology captures the state differences between nodes within the system and dynamically adjusts weights, allowing coupled nonlinear systems to maintain a stable state of synchronization after appropriate weight adjustments. To enhance synchronization optimization, we use the Master Stability Function (MSF) to demonstrate how DLS effectively adjusts networks into their synchronization regions. We introduce several variants of the DLS technique, including adaptive, supervised, and hybrid methods, effectively promoting synchronization in heterogeneous networks such as small-world, scale-free, and random networks. The efficacy of this technique is validated through its application to simple FitzHugh-Nagumo neural networks and complex Hodgkin-Huxley neuronal networks, examining its impact on both global and local synchronization. The DLS technique proposed in this study offers a new solution to synchronization problems in dynamic network environments, addressing the deficiencies in adaptability and flexibility of existing technologies and providing a fresh perspective for understanding and implementing synchronization phenomena in coupled nonlinear systems., Comment: 36 pages, 14 figures

Recently, the Floquet ${\rm Na_3Bi}$-type material has been proposed as an ideal platform for realizing various phases, i.e., the spin-degenerate Dirac semimetal (DSM) can be turned into the Weyl semimetal (WSM), and even to the Weyl half-metal (WHM). Instead of the conventional electrical methods, we use the RKKY interaction to characterize the topological phase transitions in this paper. It is found that detecting the Ising term $J_I$ is feasible for distinguishing the phase transition of DSM/WSM, since the emergence of $J_I$ is induced by the broken spin degeneracy. For the case with impurities deposited on $z$ axis (the line connecting the Weyl points), the Heisenberg term $J_H$ coexists with $J_I$ in the WSM, while $J_H$ is filtered out and only $J_I$ survives in the WHM. This magnetic filtering effect is a reflection of the fully spin-polarized property (one spin band is in the WSM phase while the other is gapped) of the WHM, and it can act a signal to capture the phase transition of WSM/WHM. This signal can not be disturbed unless the direction of the impurities greatly deviates from $z$ axis. Interestingly, as the impurities are moved into the $x$-$y$ plane, there arises another signal (a dip structure for $J_H$ at the phase boundary), which can also identify the phase transition of WSM/WHM. Furthermore, we have verified that all magnetic signals are robust to the term that breaks the electron-hole symmetry. Besides characterizing the phase transitions, our results also suggest that the Floquet DSMs are power platforms for controlling the magnetic interaction., Comment: 15 pages, 10 figures

Information based thermodynamic logic is revisited. It consists of two parts: Part A applies the modern theory of probability in which an arbitrary convex function \phi is employed as an analytic "device" to express information as statistical dependency contained in the topological sub-\sigma-algebra structure. Via thermo-doubling, Fenchel-Young equality (FYE) that consists of \phi(x) and its conjugate \psi(y) establishes the notion of equilibrium between x and y through duality symmetry and the principle of maximum entropy/minimum free energy. Part B deals with a given set of repetitive measurements, where an inherent convex function emerges via the mathematics of large deviations. Logarithm-based Shannon entropy with \phi(x)=-\log x figures prominently for i.i.d. sample statistics. Information can be a measure of the agreement between a statistical observation and its theoretical models. Maximum likelihood principle arises here and FYE provides a thermodynamic energetic narrative of recurrent data., Comment: 20 pages

Cereal grain plays a crucial role in the human diet as a major source of essential nutrients. Grain Appearance Inspection (GAI) serves as an essential process to determine grain quality and facilitate grain circulation and processing. However, GAI is routinely performed manually by inspectors with cumbersome procedures, which poses a significant bottleneck in smart agriculture. In this paper, we endeavor to develop an automated GAI system:AI4GrainInsp. By analyzing the distinctive characteristics of grain kernels, we formulate GAI as a ubiquitous problem: Anomaly Detection (AD), in which healthy and edible kernels are considered normal samples while damaged grains or unknown objects are regarded as anomalies. We further propose an AD model, called AD-GAI, which is trained using only normal samples yet can identify anomalies during inference. Moreover, we customize a prototype device for data acquisition and create a large-scale dataset including 220K high-quality images of wheat and maize kernels. Through extensive experiments, AD-GAI achieves considerable performance in comparison with advanced AD methods, and AI4GrainInsp has highly consistent performance compared to human experts and excels at inspection efficiency over 20x speedup. The dataset, code and models will be released at https://github.com/hellodfan/AI4GrainInsp., Comment: Accepted by ECAI2023. https://github.com/hellodfan/AI4GrainInsp

Dynamic shape computations have become critical in modern machine learning workloads, especially in emerging large language models. The success of these models has driven demand for deploying them to a diverse set of backend environments. In this paper, we present Relax, a compiler abstraction for optimizing end-to-end dynamic machine learning workloads. Relax introduces first-class symbolic shape annotations to track dynamic shape computations globally across the program. It also introduces a cross-level abstraction that encapsulates computational graphs, loop-level tensor programs, and library calls in a single representation to enable cross-level optimizations. We build an end-to-end compilation framework using the proposed approach to optimize dynamic shape models. Experimental results on large language models show that Relax delivers performance competitive with state-of-the-art hand-optimized systems across platforms and enables deployment of emerging dynamic models to a broader set of environments, including mobile phones, embedded devices, and web browsers.

Assessing causal effects in the presence of unobserved confounding is a challenging problem. Existing studies leveraged proxy variables or multiple treatments to adjust for the confounding bias. In particular, the latter approach attributes the impact on a single outcome to multiple treatments, allowing estimating latent variables for confounding control. Nevertheless, these methods primarily focus on a single outcome, whereas in many real-world scenarios, there is greater interest in studying the effects on multiple outcomes. Besides, these outcomes are often coupled with multiple treatments. Examples include the intensive care unit (ICU), where health providers evaluate the effectiveness of therapies on multiple health indicators. To accommodate these scenarios, we consider a new setting dubbed as multiple treatments and multiple outcomes. We then show that parallel studies of multiple outcomes involved in this setting can assist each other in causal identification, in the sense that we can exploit other treatments and outcomes as proxies for each treatment effect under study. We proceed with a causal discovery method that can effectively identify such proxies for causal estimation. The utility of our method is demonstrated in synthetic data and sepsis disease., Comment: Preprint, under review

Proximal causal learning is a promising framework for identifying the causal effect under the existence of unmeasured confounders. Within this framework, the doubly robust (DR) estimator was derived and has shown its effectiveness in estimation, especially when the model assumption is violated. However, the current form of the DR estimator is restricted to binary treatments, while the treatment can be continuous in many real-world applications. The primary obstacle to continuous treatments resides in the delta function present in the original DR estimator, making it infeasible in causal effect estimation and introducing a heavy computational burden in nuisance function estimation. To address these challenges, we propose a kernel-based DR estimator that can well handle continuous treatments. Equipped with its smoothness, we show that its oracle form is a consistent approximation of the influence function. Further, we propose a new approach to efficiently solve the nuisance functions. We then provide a comprehensive convergence analysis in terms of the mean square error. We demonstrate the utility of our estimator on synthetic datasets and real-world applications., Comment: Published as a conference paper at ICLR 2024

With the rapid development of electronic science and technology, the research on wearable devices is constantly updated, but for now, it is not comprehensive for wearable devices to recognize and analyze the movement of specific sports. Based on this, this paper improves wearable devices of table tennis sport, and realizes the pattern recognition and evaluation of table tennis players' motor skills through artificial intelligence. Firstly, a device is designed to collect the movement information of table tennis players and the actual movement data is processed. Secondly, a sliding window is made to divide the collected motion data into a characteristic database of six table tennis benchmark movements. Thirdly, motion features were constructed based on feature engineering, and motor skills were identified for different models after dimensionality reduction. Finally, the hierarchical evaluation system of motor skills is established with the loss functions of different evaluation indexes. The results show that in the recognition of table tennis players' motor skills, the feature-based BP neural network proposed in this paper has higher recognition accuracy and stronger generalization ability than the traditional convolutional neural network., Comment: 34pages, 16figures

The non-local thermodynamical equilibrium (NLTE) line formation of Y I and Y II is considered in 1D LTE model atmospheres of F-G-K-type stars. The model atom was constructed with the most up-to-date atomic data, including quantum cross sections and rate coefficients for transitions in inelastic collisions of Y I and Y II with hydrogen atoms. For seven reference stars, we obtained an agreement between NLTE abundances inferred from the two ionization stages, while the difference in LTE abundance (Y I - Y II) can reach up to -0.31 dex. In the atmospheres of F-G-K-type stars, for both Y I and Y II lines, the NLTE abundance corrections are positive. In solar metallicity stars, the NLTE abundance corrections for Y II lines do not exceed 0.12 dex, while in atmospheres of metal-poor stars they do not exceed 0.21 dex. For Y I lines, the NLTE abundance corrections can reach up to 0.5 dex. We determined the yttrium NLTE abundances for a sample of 65 F and G dwarfs and subgiants in the -2.62~$\leq$~[Fe/H]~$\leq$~+0.24 metallicity range, using high-resolution spectra. For stars with [Fe/H]~$\leq$~-1.5, [Y/Fe] versus [Fe/H] diagram reveals positive trend with an average value of [Y/Fe]~$\simeq$~0. For metal-poor stars, among Sr, Y, and Zr, the arrangement [Sr/Fe] < [Y/Fe] < [Zr/Fe] remains consistent. The current study is useful for the Galactic chemical evolution research. The model atom will be applied for NLTE yttrium abundance determination in very metal-poor stars studied with LAMOST and Subaru., Comment: 13 figures, 8 tables, accepted to ApJ

Disorder-induced heating (DIH) prevents ultracold neutral plasma into electron strong coupling regime. Here we propose a scheme to suppress electronic DIH via optical lattice. We simulate the evolution dynamics of ultracold neutral plasma constrained by three-dimensional optical lattice using classical molecular dynamics method. The results show that for experimentally achievable condition, electronic DIH is suppressed by a factor of 1.3, and the Coulomb coupling strength can reach to 0.8 which is approaching the strong coupling regime. Suppressing electronic DIH via optical lattice may pave a way for the research of electronic strongly coupled plasma.

Utilizing a simplified quantum model approach, the low-energy inelastic collision processes between yttrium atoms (ions) and hydrogen atoms have been studied. Rate coefficients corresponding to the mutual neutralization, ion-pair formation, excitation, and de-excitation processes for the above collision systems have been provided in the temperature range of 1000-10000K. 3 ionic states and 73 covalent states are considered in calculations for the collisions of yttrium atoms with hydrogen atoms, which include 6 molecular symmetries and 4074 partial inelastic reaction processes. For the collisions of yttrium ions with hydrogen atoms, 1 ionic state and 116 covalent states are included, which related to 3 molecular symmetries and 13572 partial inelastic collision processes. It is found that the rate coefficients for the mutual neutralization process have a maximum at T = 6000K, which is an order of magnitude higher than those of other processes. Notably, the positions of optimal windows for the collisions of yttrium atoms and ions with hydrogen atoms are found near electronic binding energy -2eV (Y) and -4.4eV (Y$^+$), respectively. The scattering channels located in or near these optimal windows have intermediate-to-large rate coefficients (greater than $10^{-12}$ cm$^3$s$^{-1}$). The reported data should be useful in the study of non-local thermodynamic equilibrium modeling., Comment: 15 pages, 6 figures, 3 tables

Despite the tremendous advances achieved over the past years by deep learning techniques, the latest risk prediction models for industrial applications still rely on highly handtuned stage-wised statistical learning tools, such as gradient boosting and random forest methods. Different from images or languages, real-world financial data are high-dimensional, sparse, noisy and extremely imbalanced, which makes deep neural network models particularly challenging to train and fragile in practice. In this work, we propose DeRisk, an effective deep learning risk prediction framework for credit risk prediction on real-world financial data. DeRisk is the first deep risk prediction model that outperforms statistical learning approaches deployed in our company's production system. We also perform extensive ablation studies on our method to present the most critical factors for the empirical success of DeRisk.