arXiv subjects

Yi Yao

Publications and source records attributed to Yi Yao.

At least 55 records · Page 3Linked to original sources

arXiv2022

Relative Ding Stability and an Obstruction to the Existence of Mabuchi Solitons

Mabuchi solitons generalize Kähler-Einstein metrics on Fano manifolds, which constitute a Yau-Tian-Donaldson type correspondence with relative Ding stability. Comparing with Kähler-Ricci solitons, there is a distinct necessary condition for the existence. We show this condition can be implied by the uniformly relative Ding stability. For this we study the inner product of $\mathbb{C}^{*}$-actions on equivariant test-configurations and obtain an integration formula over the total space. To analyze the uniform stability, by adapting Okounkov body construction to the setting of torus action, we give a convex-geometry description for the reduced non-Archimedean J-functionals.

math.DG↗

arXiv2021

Machine Learning Applications on Neuroimaging for Diagnosis and Prognosis of Epilepsy: A Review

Machine learning is playing an increasingly important role in medical image analysis, spawning new advances in the clinical application of neuroimaging. There have been some reviews on machine learning and epilepsy before, and they mainly focused on electrophysiological signals such as electroencephalography (EEG) and stereo electroencephalography (SEEG), while neglecting the potential of neuroimaging in epilepsy research. Neuroimaging has its important advantages in confirming the range of the epileptic region, which is essential in presurgical evaluation and assessment after surgery. However, it is difficult for EEG to locate the accurate epilepsy lesion region in the brain. In this review, we emphasize the interaction between neuroimaging and machine learning in the context of epilepsy diagnosis and prognosis. We start with an overview of epilepsy and typical neuroimaging modalities used in epilepsy clinics, MRI, DWI, fMRI, and PET. Then, we elaborate two approaches in applying machine learning methods to neuroimaging data: i) the conventional machine learning approach combining manual feature engineering and classifiers, ii) the deep learning approach, such as the convolutional neural networks and autoencoders. Subsequently, the application of machine learning on epilepsy neuroimaging, such as segmentation, localization, and lateralization tasks, as well as tasks directly related to diagnosis and prognosis are looked into in detail. Finally, we discuss the current achievements, challenges, and potential future directions in this field, hoping to pave the way for computer-aided diagnosis and prognosis of epilepsy.

cs.LG↗

arXiv2021

Generating and Evaluating Explanations of Attended and Error-Inducing Input Regions for VQA Models

Attention maps, a popular heatmap-based explanation method for Visual Question Answering (VQA), are supposed to help users understand the model by highlighting portions of the image/question used by the model to infer answers. However, we see that users are often misled by current attention map visualizations that point to relevant regions despite the model producing an incorrect answer. Hence, we propose Error Maps that clarify the error by highlighting image regions where the model is prone to err. Error maps can indicate when a correctly attended region may be processed incorrectly leading to an incorrect answer, and hence, improve users' understanding of those cases. To evaluate our new explanations, we further introduce a metric that simulates users' interpretation of explanations to evaluate their potential helpfulness to understand model correctness. We finally conduct user studies to see that our new explanations help users understand model correctness better than baselines by an expected 30\% and that our proxy helpfulness metrics correlate strongly ($ρ>0.97$) with how well users can predict model correctness.

cs.CV↗

arXiv2021

Improving Users' Mental Model with Attention-directed Counterfactual Edits

In the domain of Visual Question Answering (VQA), studies have shown improvement in users' mental model of the VQA system when they are exposed to examples of how these systems answer certain Image-Question (IQ) pairs. In this work, we show that showing controlled counterfactual image-question examples are more effective at improving the mental model of users as compared to simply showing random examples. We compare a generative approach and a retrieval-based approach to show counterfactual examples. We use recent advances in generative adversarial networks (GANs) to generate counterfactual images by deleting and inpainting certain regions of interest in the image. We then expose users to changes in the VQA system's answer on those altered images. To select the region of interest for inpainting, we experiment with using both human-annotated attention maps and a fully automatic method that uses the VQA system's attention values. Finally, we test the user's mental model by asking them to predict the model's performance on a test counterfactual image. We note an overall improvement in users' accuracy to predict answer change when shown counterfactual explanations. While realistic retrieved counterfactuals obviously are the most effective at improving the mental model, we show that a generative approach can also be equally effective.

cs.CV↗

arXiv2021

Mabuchi Solitons and Relative Ding Stability of Toric Fano Varieties

As a generalization of Kahler-Einstein metrics for Fano manifolds with nonvanishing Futaki invariant, Mabuchi solitons are critical points of a Calabi-type energy functional. We study their existence on toric Fano varieties and the underlying algebraic stability notion: relative Ding stability. As a toy model for a YTD type correspondence, a new feature is the emergence of a non-uniformly stable case. We show a partial coercivity for the modified Ding functionals in this case, and obtain singular Mabuchi solitons via a variational approach. In the unstable case, we determine the maximal destabilizer which is a simple convex function over the moment polytope, and establish a Moment-Weight equality which connects the infimum of a Calabi-type energy and the Berman-Ding invariant.

math.DG↗

arXiv2021

Confidence Calibration for Domain Generalization under Covariate Shift

Existing calibration algorithms address the problem of covariate shift via unsupervised domain adaptation. However, these methods suffer from the following limitations: 1) they require unlabeled data from the target domain, which may not be available at the stage of calibration in real-world applications and 2) their performance depends heavily on the disparity between the distributions of the source and target domains. To address these two limitations, we present novel calibration solutions via domain generalization. Our core idea is to leverage multiple calibration domains to reduce the effective distribution disparity between the target and calibration domains for improved calibration transfer without needing any data from the target domain. We provide theoretical justification and empirical experimental results to demonstrate the effectiveness of our proposed algorithms. Compared against state-of-the-art calibration methods designed for domain adaptation, we observe a decrease of 8.86 percentage points in expected calibration error or, equivalently, an increase of 35 percentage points in improvement ratio for multi-class classification on the Office-Home dataset.

cs.LG↗

arXiv2021

Modular Adaptation for Cross-Domain Few-Shot Learning

Adapting pre-trained representations has become the go-to recipe for learning new downstream tasks with limited examples. While literature has demonstrated great successes via representation learning, in this work, we show that substantial performance improvement of downstream tasks can also be achieved by appropriate designs of the adaptation process. Specifically, we propose a modular adaptation method that selectively performs multiple state-of-the-art (SOTA) adaptation methods in sequence. As different downstream tasks may require different types of adaptation, our modular adaptation enables the dynamic configuration of the most suitable modules based on the downstream task. Moreover, as an extension to existing cross-domain 5-way k-shot benchmarks (e.g., miniImageNet -> CUB), we create a new high-way (~100) k-shot benchmark with data from 10 different datasets. This benchmark provides a diverse set of domains and allows the use of stronger representations learned from ImageNet. Experimental results show that by customizing adaptation process towards downstream tasks, our modular adaptation pipeline (MAP) improves 3.1% in 5-shot classification accuracy over baselines of finetuning and Prototypical Networks.

cs.CV↗

arXiv2021

All-electron periodic $G_0W_0$ implementation with numerical atomic orbital basis functions: algorithm and benchmarks

We present an all-electron, periodic {\GnWn} implementation within the numerical atomic orbital (NAO) basis framework. A localized variant of the resolution-of-the-identity (RI) approximation is employed to significantly reduce the computational cost of evaluating and storing the two-electron Coulomb repulsion integrals. We demonstrate that the error arising from localized RI approximation can be reduced to an insignificant level by enhancing the set of auxiliary basis functions, used to expand the products of two single-particle NAOs. An efficient algorithm is introduced to deal with the Coulomb singularity in the Brillouin zone sampling that is suitable for the NAO framework. We perform systematic convergence tests and identify a set of computational parameters, which can serve as the default choice for most practical purposes. Benchmark calculations are carried out for a set of prototypical semiconductors and insulators, and compared to independent reference values obtained from an independent $G_0W_0$ implementation based on linearized augmented plane waves (LAPW) plus high-energy localized orbitals (HLOs) basis set, as well as experimental results. With a moderate (FHI-aims \textit{tier} 2) NAO basis set, our $G_0W_0$ calculations produce band gaps that typically lie in between the standard LAPW and the LAPW+HLO results. Complementing \textit{tier} 2 with highly localized Slater-type orbitals (STOs), we find that the obtained band gaps show an overall convergence towards the LAPW+HLO results. The algorithms and techniques developed in this work pave the way for efficient implementations of correlated methods within the NAO framework.

cond-mat.mtrl-sci↗

arXiv2020

Hybrid Consistency Training with Prototype Adaptation for Few-Shot Learning

Few-Shot Learning (FSL) aims to improve a model's generalization capability in low data regimes. Recent FSL works have made steady progress via metric learning, meta learning, representation learning, etc. However, FSL remains challenging due to the following longstanding difficulties. 1) The seen and unseen classes are disjoint, resulting in a distribution shift between training and testing. 2) During testing, labeled data of previously unseen classes is sparse, making it difficult to reliably extrapolate from labeled support examples to unlabeled query examples. To tackle the first challenge, we introduce Hybrid Consistency Training to jointly leverage interpolation consistency, including interpolating hidden features, that imposes linear behavior locally and data augmentation consistency that learns robust embeddings against sample variations. As for the second challenge, we use unlabeled examples to iteratively normalize features and adapt prototypes, as opposed to commonly used one-time update, for more reliable prototype-based transductive inference. We show that our method generates a 2% to 5% improvement over the state-of-the-art methods with similar backbones on five FSL datasets and, more notably, a 7% to 8% improvement for more challenging cross-domain FSL.

cs.CV↗

arXiv2020

The Impact of Explanations on AI Competency Prediction in VQA

Explainability is one of the key elements for building trust in AI systems. Among numerous attempts to make AI explainable, quantifying the effect of explanations remains a challenge in conducting human-AI collaborative tasks. Aside from the ability to predict the overall behavior of AI, in many applications, users need to understand an AI agent's competency in different aspects of the task domain. In this paper, we evaluate the impact of explanations on the user's mental model of AI agent competency within the task of visual question answering (VQA). We quantify users' understanding of competency, based on the correlation between the actual system performance and user rankings. We introduce an explainable VQA system that uses spatial and object features and is powered by the BERT language model. Each group of users sees only one kind of explanation to rank the competencies of the VQA model. The proposed model is evaluated through between-subject experiments to probe explanations' impact on the user's perception of competency. The comparison between two VQA models shows BERT based explanations and the use of object features improve the user's prediction of the model's competencies.

cs.CV↗

arXiv2020

Progressive Growing of Neural ODEs

Neural Ordinary Differential Equations (NODEs) have proven to be a powerful modeling tool for approximating (interpolation) and forecasting (extrapolation) irregularly sampled time series data. However, their performance degrades substantially when applied to real-world data, especially long-term data with complex behaviors (e.g., long-term trend across years, mid-term seasonality across months, and short-term local variation across days). To address the modeling of such complex data with different behaviors at different frequencies (time spans), we propose a novel progressive learning paradigm of NODEs for long-term time series forecasting. Specifically, following the principle of curriculum learning, we gradually increase the complexity of data and network capacity as training progresses. Our experiments with both synthetic data and real traffic data (PeMS Bay Area traffic data) show that our training methodology consistently improves the performance of vanilla NODEs by over 64%.

cs.LG↗

arXiv2020

A Study on Multimodal and Interactive Explanations for Visual Question Answering

Explainability and interpretability of AI models is an essential factor affecting the safety of AI. While various explainable AI (XAI) approaches aim at mitigating the lack of transparency in deep networks, the evidence of the effectiveness of these approaches in improving usability, trust, and understanding of AI systems are still missing. We evaluate multimodal explanations in the setting of a Visual Question Answering (VQA) task, by asking users to predict the response accuracy of a VQA agent with and without explanations. We use between-subjects and within-subjects experiments to probe explanation effectiveness in terms of improving user prediction accuracy, confidence, and reliance, among other factors. The results indicate that the explanations help improve human prediction accuracy, especially in trials when the VQA system's answer is inaccurate. Furthermore, we introduce active attention, a novel method for evaluating causal attentional effects through intervention by editing attention maps. User explanation ratings are strongly correlated with human prediction accuracy and suggest the efficacy of these explanations in human-machine AI collaboration tasks.

cs.AI↗

arXiv2019

Can You Explain That? Lucid Explanations Help Human-AI Collaborative Image Retrieval

While there have been many proposals on making AI algorithms explainable, few have attempted to evaluate the impact of AI-generated explanations on human performance in conducting human-AI collaborative tasks. To bridge the gap, we propose a Twenty-Questions style collaborative image retrieval game, Explanation-assisted Guess Which (ExAG), as a method of evaluating the efficacy of explanations (visual evidence or textual justification) in the context of Visual Question Answering (VQA). In our proposed ExAG, a human user needs to guess a secret image picked by the VQA agent by asking natural language questions to it. We show that overall, when AI explains its answers, users succeed more often in guessing the secret image correctly. Notably, a few correct explanations can readily improve human performance when VQA answers are mostly incorrect as compared to no-explanation games. Furthermore, we also show that while explanations rated as "helpful" significantly improve human performance, "incorrect" and "unhelpful" explanations can degrade performance as compared to no-explanation games. Our experiments, therefore, demonstrate that ExAG is an effective means to evaluate the efficacy of AI-generated explanations on a human-AI collaborative task.

cs.CY↗

arXiv2019

K-shell Core Electron Excitations in the Electronic Stopping of Protons in Water

Developing a detailed understanding for the role of core electron excitation in liquid water under proton irradiation has become important due to the growing use of proton beams in radiation oncology. Using a first-principles, non-equilibrium simulation approach based on real-time time-dependent density functional theory, we determine the electronic stopping power, the velocity-dependent energy transfer rate from irradiating ions to electrons. The electronic stopping power curve agrees quantitatively over the entire velocity range for which experimental data is available. Also notably, we observe significant differences between our first-principles results and commonly-used perturbation theoretic models. Excitations of the water molecules' oxygen core electrons are a crucial factor in determining the electronic stopping power curve beyond its maximum. The core electron contribution is responsible for as much as one-third of the stopping power at the high proton velocity of 8.0 a.u. (1.6 MeV). These K-shell core electron excitations not only provide an additional channel for the energy transfer but they also significantly influence the valence electron excitation. In the excitation process, generated holes remain highly localized within a few angstroms around the irradiating proton path whereas electrons are excited away from the path. In spite of its great contribution to the electronic stopping power, the K-shell electrons play a rather minor role in terms of the excitation density: Only one percent of the hole population comprises K-shell holes, even at the high proton velocity of 8.0 a.u.. The excitation behavior revealed here is distinctly different from that of X/gamma-ray photon radiation, which is the most commonly used type of ionizing radiation in radiation oncology.

cond-mat.mtrl-sci↗

arXiv2019

Resonant inelastic x-ray scattering of magnetic excitations under pressure

Resonant inelastic x-ray scattering (RIXS) is an extremely valuable tool for the study of elementary, including magnetic, excitations in matter. Latest developments of this technique mostly aimed at improving the energy resolution and performing polarization analysis of the scattered radiation, with a great impact on the interpretation and applicability of RIXS. Instead, this article focuses on the sample environment and presents a setup for high-pressure low-temperature RIXS measurements of low-energy excitations. The feasibility of these experiments is proved by probing the magnetic excitations of the bilayer iridate Sr$_3$Ir$_2$O$_7$ at pressures up to 12 GPa.

cond-mat.str-el↗

arXiv2019

Stacked Spatio-Temporal Graph Convolutional Networks for Action Segmentation

We propose novel Stacked Spatio-Temporal Graph Convolutional Networks (Stacked-STGCN) for action segmentation, i.e., predicting and localizing a sequence of actions over long videos. We extend the Spatio-Temporal Graph Convolutional Network (STGCN) originally proposed for skeleton-based action recognition to enable nodes with different characteristics (e.g., scene, actor, object, action, etc.), feature descriptors with varied lengths, and arbitrary temporal edge connections to account for large graph deformation commonly associated with complex activities. We further introduce the stacked hourglass architecture to STGCN to leverage the advantages of an encoder-decoder design for improved generalization performance and localization accuracy. We explore various descriptors such as frame-level VGG, segment-level I3D, RCNN-based object, etc. as node descriptors to enable action segmentation based on joint inference over comprehensive contextual information. We show results on CAD120 (which provides pre-computed node features and edge weights for fair performance comparison across algorithms) as well as a more complex real-world activity dataset, Charades. Our Stacked-STGCN in general achieves 4.0% performance improvement over the best reported results in F1 score on CAD120 and 1.3% in mAP on Charades using VGG features.

cs.CV↗

arXiv2019

Propagation of maximally localized Wannier functions in real-time TDDFT

Real-time, time-dependent density functional theory (RT-TDDFT) has gained popularity as a first-principles approach to study a variety of excited-state phenomena such as optical excitations and electronic stopping. Within RT-TDDFT simulations, the gauge freedom of the time-dependent electronic orbitals can be exploited for numerical and scientific convenience while the unitary transformation does not alter physical properties calculated from the quantum dynamics of electrons. Exploiting this gauge freedom, we demonstrate propagation of maximally localized Wannier functions within RT-TDDFT. We illustrate its great utility through a number of examples including its application to optical excitation in extended systems using the so-called length gauge, interpreting electronic stopping excitation, and simulating electric field-driven quantized charge transport. We implemented the approach within our plane-wave pseudopotential RT-TDDFT module of the QB@LL code, and performance of the implementation is also discussed.

cond-mat.mtrl-sci↗

arXiv2018

Examining real-time TDDFT non-equilibrium simulations for the calculation of electronic stopping power

In ion irradiation processes, electronic stopping power describes the energy transfer rate from the irradiating ion to the target material's electrons. Due to the scarcity and significant uncertainties in experimental electronic stopping power data for materials beyond simple solids, there has been growing interest in the use of first-principles theory for calculating electronic stopping power. In recent years, advances in high-performance computing have opened the door to fully first-principles nonequilibrium simulations based on real-time time-dependent density functional theory (RT-TDDFT). While it has been demonstrated that the RT-TDDFT approach is capable of predicting electronic stopping power for a wide range of condensed matter systems, there has yet to be an exhaustive examination of the physical and numerical approximations involved and their effects on the calculated stopping power. We discuss the results of such a study for crystalline silicon with protons as irradiating ions. We examine the influences of key approximations in RT-TDDFT nonequilibrium simulations on the calculated electronic stopping power, including approximations related to basis sets, finite size effects, exchange-correlation approximation, pseudopotentials, and more. Finally, we propose a simple and efficient correction scheme to account for the contribution from core-electron excitations to the stopping power, as it was found to be significant for large proton velocities.

cond-mat.mtrl-sci↗