arXiv subjects

Gen Li

Publications and source records attributed to Gen Li.

At least 235 records · Page 13Linked to original sources

arXiv2021

SuperStyleNet: Deep Image Synthesis with Superpixel Based Style Encoder

Existing methods for image synthesis utilized a style encoder based on stacks of convolutions and pooling layers to generate style codes from input images. However, the encoded vectors do not necessarily contain local information of the corresponding images since small-scale objects are tended to "wash away" through such downscaling procedures. In this paper, we propose deep image synthesis with superpixel based style encoder, named as SuperStyleNet. First, we directly extract the style codes from the original image based on superpixels to consider local objects. Second, we recover spatial relationships in vectorized style codes based on graphical analysis. Thus, the proposed network achieves high-quality image synthesis by mapping the style codes into semantic labels. Experimental results show that the proposed method outperforms state-of-the-art ones in terms of visual quality and quantitative measurements. Furthermore, we achieve elaborate spatial style editing by adjusting style codes.

cs.CV↗

arXiv2021

Sample-Efficient Reinforcement Learning Is Feasible for Linearly Realizable MDPs with Limited Revisiting

Low-complexity models such as linear function representation play a pivotal role in enabling sample-efficient reinforcement learning (RL). The current paper pertains to a scenario with value-based linear representation, which postulates the linear realizability of the optimal Q-function (also called the "linear $Q^{\star}$ problem"). While linear realizability alone does not allow for sample-efficient solutions in general, the presence of a large sub-optimality gap is a potential game changer, depending on the sampling mechanism in use. Informally, sample efficiency is achievable with a large sub-optimality gap when a generative model is available but is unfortunately infeasible when we turn to standard online RL settings. In this paper, we make progress towards understanding this linear $Q^{\star}$ problem by investigating a new sampling protocol, which draws samples in an online/exploratory fashion but allows one to backtrack and revisit previous states in a controlled and infrequent manner. This protocol is more flexible than the standard online RL setting, while being practically relevant and far more restrictive than the generative model. We develop an algorithm tailored to this setting, achieving a sample complexity that scales polynomially with the feature dimension, the horizon, and the inverse sub-optimality gap, but not the size of the state/action space. Our findings underscore the fundamental interplay between sampling protocols and low-complexity structural representation in RL.

cs.LG↗

arXiv2021

Sample Complexity of Asynchronous Q-Learning: Sharper Analysis and Variance Reduction

Asynchronous Q-learning aims to learn the optimal action-value function (or Q-function) of a Markov decision process (MDP), based on a single trajectory of Markovian samples induced by a behavior policy. Focusing on a $γ$-discounted MDP with state space $\mathcal{S}$ and action space $\mathcal{A}$, we demonstrate that the $\ell_{\infty}$-based sample complexity of classical asynchronous Q-learning --- namely, the number of samples needed to yield an entrywise $\varepsilon$-accurate estimate of the Q-function --- is at most on the order of $\frac{1}{μ_{\min}(1-γ)^5\varepsilon^2}+ \frac{t_{mix}}{μ_{\min}(1-γ)}$ up to some logarithmic factor, provided that a proper constant learning rate is adopted. Here, $t_{mix}$ and $μ_{\min}$ denote respectively the mixing time and the minimum state-action occupancy probability of the sample trajectory. The first term of this bound matches the sample complexity in the synchronous case with independent samples drawn from the stationary distribution of the trajectory. The second term reflects the cost taken for the empirical distribution of the Markovian trajectory to reach a steady state, which is incurred at the very beginning and becomes amortized as the algorithm runs. Encouragingly, the above bound improves upon the state-of-the-art result \cite{qu2020finite} by a factor of at least $|\mathcal{S}||\mathcal{A}|$ for all scenarios, and by a factor of at least $t_{mix}|\mathcal{S}||\mathcal{A}|$ for any sufficiently small accuracy level $\varepsilon$. Further, we demonstrate that the scaling on the effective horizon $\frac{1}{1-γ}$ can be improved by means of variance reduction.

cs.LG↗

arXiv2021

Nonconvex Low-Rank Tensor Completion from Noisy Data

We study a noisy tensor completion problem of broad practical interest, namely, the reconstruction of a low-rank tensor from highly incomplete and randomly corrupted observations of its entries. While a variety of prior work has been dedicated to this problem, prior algorithms either are computationally too expensive for large-scale applications, or come with sub-optimal statistical guarantees. Focusing on "incoherent" and well-conditioned tensors of a constant CP rank, we propose a two-stage nonconvex algorithm -- (vanilla) gradient descent following a rough initialization -- that achieves the best of both worlds. Specifically, the proposed nonconvex algorithm faithfully completes the tensor and retrieves all individual tensor factors within nearly linear time, while at the same time enjoying near-optimal statistical guarantees (i.e. minimal sample complexity and optimal estimation accuracy). The estimation errors are evenly spread out across all entries, thus achieving optimal $\ell_{\infty}$ statistical accuracy. We have also discussed how to extend our approach to accommodate asymmetric tensors. The insight conveyed through our analysis of nonconvex optimization might have implications for other tensor estimation problems.

cs.LG↗

arXiv2021

Adaptive Prototype Learning and Allocation for Few-Shot Segmentation

Prototype learning is extensively used for few-shot segmentation. Typically, a single prototype is obtained from the support feature by averaging the global object information. However, using one prototype to represent all the information may lead to ambiguities. In this paper, we propose two novel modules, named superpixel-guided clustering (SGC) and guided prototype allocation (GPA), for multiple prototype extraction and allocation. Specifically, SGC is a parameter-free and training-free approach, which extracts more representative prototypes by aggregating similar feature vectors, while GPA is able to select matched prototypes to provide more accurate guidance. By integrating the SGC and GPA together, we propose the Adaptive Superpixel-guided Network (ASGNet), which is a lightweight model and adapts to object scale and shape variation. In addition, our network can easily generalize to k-shot segmentation with substantial improvement and no additional computational cost. In particular, our evaluations on COCO demonstrate that ASGNet surpasses the state-of-the-art method by 5% in 5-shot segmentation.

cs.CV↗

arXiv2021

Edge and Identity Preserving Network for Face Super-Resolution

Face super-resolution (SR) has become an indispensable function in security solutions such as video surveillance and identification system, but the distortion in facial components is a great challenge in it. Most state-of-the-art methods have utilized facial priors with deep neural networks. These methods require extra labels, longer training time, and larger computation memory. In this paper, we propose a novel Edge and Identity Preserving Network for Face SR Network, named as EIPNet, to minimize the distortion by utilizing a lightweight edge block and identity information. We present an edge block to extract perceptual edge information, and concatenate it to the original feature maps in multiple scales. This structure progressively provides edge information in reconstruction to aggregate local and global structural information. Moreover, we define an identity loss function to preserve identification of SR images. The identity loss function compares feature distributions between SR images and their ground truth to recover identities in SR images. In addition, we provide a luminance-chrominance error (LCE) to separately infer brightness and color information in SR images. The LCE method not only reduces the dependency of color information by dividing brightness and color components but also enables our network to reflect differences between SR images and their ground truth in two color spaces of RGB and YUV. The proposed method facilitates the proposed SR network to elaborately restore facial components and generate high quality 8x scaled SR images with a lightweight network structure. Furthermore, our network is able to reconstruct an 128x128 SR image with 215 fps on a GTX 1080Ti GPU. Extensive experiments demonstrate that our network qualitatively and quantitatively outperforms state-of-the-art methods on two challenging datasets: CelebA and VGGFace2.

cs.CV↗

arXiv2020

Subspace Estimation from Unbalanced and Incomplete Data Matrices: $\ell_{2,\infty}$ Statistical Guarantees

This paper is concerned with estimating the column space of an unknown low-rank matrix $\boldsymbol{A}^{\star}\in\mathbb{R}^{d_{1}\times d_{2}}$, given noisy and partial observations of its entries. There is no shortage of scenarios where the observations -- while being too noisy to support faithful recovery of the entire matrix -- still convey sufficient information to enable reliable estimation of the column space of interest. This is particularly evident and crucial for the highly unbalanced case where the column dimension $d_{2}$ far exceeds the row dimension $d_{1}$, which is the focal point of the current paper. We investigate an efficient spectral method, which operates upon the sample Gram matrix with diagonal deletion. While this algorithmic idea has been studied before, we establish new statistical guarantees for this method in terms of both $\ell_{2}$ and $\ell_{2,\infty}$ estimation accuracy, which improve upon prior results if $d_{2}$ is substantially larger than $d_{1}$. To illustrate the effectiveness of our findings, we derive matching minimax lower bounds with respect to the noise levels, and develop consequences of our general theory for three applications of practical importance: (1) tensor completion from noisy data, (2) covariance estimation / principal component analysis with missing data, and (3) community recovery in bipartite graphs. Our theory leads to improved performance guarantees for all three cases.

math.ST↗

arXiv2020

It's All Relative: New Regression Paradigm for Microbiome Compositional Data

Microbiome data are complex in nature, involving high dimensionality, compositionally, zero inflation, and taxonomic hierarchy. Compositional data reside in a simplex that does not admit the standard Euclidean geometry. Most existing compositional regression methods rely on transformations that are inadequate or even inappropriate in modeling data with excessive zeros and taxonomic structure. We develop a novel relative-shift regression framework that directly uses compositions as predictors. The new framework provides a paradigm shift for compositional regression and offers a superior biological interpretation. New equi-sparsity and taxonomy-guided regularization methods and an efficient smoothing proximal gradient algorithm are developed to facilitate feature aggregation and dimension reduction in regression. As a result, the framework can automatically identify clinically relevant microbes even if they are important at different taxonomic levels. A unified finite-sample prediction error bound is developed for the proposed regularized estimators. We demonstrate the efficacy of the proposed methods in extensive simulation studies. The application to a preterm infant study reveals novel insights of association between the gut microbiome and neurodevelopment.

stat.ME↗

arXiv2020

The More Data, the Better? Demystifying Deletion-Based Methods in Linear Regression with Missing Data

We compare two deletion-based methods for dealing with the problem of missing observations in linear regression analysis. One is the complete-case analysis (CC, or listwise deletion) that discards all incomplete observations and only uses common samples for ordinary least-squares estimation. The other is the available-case analysis (AC, or pairwise deletion) that utilizes all available data to estimate the covariance matrices and applies these matrices to construct the normal equation. We show that the estimates from both methods are asymptotically unbiased and further compare their asymptotic variances in some typical situations. Surprisingly, using more data (i.e., AC) does not necessarily lead to better asymptotic efficiency in many scenarios. Missing patterns, covariance structure and true regression coefficient values all play a role in determining which is better. We further conduct simulation studies to corroborate the findings and demystify what has been missed or misinterpreted in the literature. Some detailed proofs and simulation results are available in the online supplemental materials.

stat.ME↗

arXiv2020

Learning Topological Motion Primitives for Knot Planning

In this paper, we approach the challenging problem of motion planning for knot tying. We propose a hierarchical approach in which the top layer produces a topological plan and the bottom layer translates this plan into continuous robot motion. The top layer decomposes a knotting task into sequences of abstract topological actions based on knot theory. The bottom layer translates each of these abstract actions into robot motion trajectories through learned topological motion primitives. To adapt each topological action to the specific rope geometry, the motion primitives take the observed rope configuration as input. We train the motion primitives by imitating human demonstrations and reinforcement learning in simulation. To generalize human demonstrations of simple knots into more complex knots, we observe similarities in the motion strategies of different topological actions and design the neural network structure to exploit such similarities. We demonstrate that our learned motion primitives can be used to efficiently generate motion plans for tying the overhand knot. The motion plan can then be executed on a real robot using visual tracking and Model Predictive Control. We also demonstrate that our learned motion primitives can be composed to tie a more complex pentagram-like knot despite being only trained on human demonstrations of simpler knots.

cs.RO↗

arXiv2020

The Efficacy of $L_1$ Regularization in Two-Layer Neural Networks

A crucial problem in neural networks is to select the most appropriate number of hidden neurons and obtain tight statistical risk bounds. In this work, we present a new perspective towards the bias-variance tradeoff in neural networks. As an alternative to selecting the number of neurons, we theoretically show that $L_1$ regularization can control the generalization error and sparsify the input dimension. In particular, with an appropriate $L_1$ regularization on the output layer, the network can produce a statistical risk that is near minimax optimal. Moreover, an appropriate $L_1$ regularization on the input layer leads to a risk bound that does not involve the input data dimension. Our analysis is based on a new amalgamation of dimension-based and norm-based complexity analysis to bound the generalization error. A consequent observation from our results is that an excessively large number of neurons do not necessarily inflate generalization errors under a suitable regularization.

cs.LG↗

arXiv2020

Multivariate Log-Contrast Regression with Sub-Compositional Predictors: Testing the Association Between Preterm Infants' Gut Microbiome and Neurobehavioral Outcomes

The so-called gut-brain axis has stimulated extensive research on microbiomes. One focus is to assess the association between certain clinical outcomes and the relative abundances of gut microbes, which can be presented as sub-compositional data in conformity with the taxonomic hierarchy of bacteria. Motivated by a study for identifying the microbes in the gut microbiome of preterm infants that impact their later neurobehavioral outcomes, we formulate a constrained integrative multi-view regression, where the neurobehavioral scores form multivariate response, the sub-compositional microbiome data form multi-view feature matrices, and a set of linear constraints on their corresponding sub-coefficient matrices ensures the conformity to the simplex geometry. To enable joint selection and inference of sub-compositions/views, we assume all the sub-coefficient matrices are possibly of low-rank, i.e., the outcomes are associated with the microbiome through different sets of latent sub-compositional factors from different taxa. We propose a scaled composite nuclear norm penalization approach for model estimation and develop a hypothesis testing procedure through de-biasing to assess the significance of different views. Simulation studies confirm the effectiveness of the proposed procedure. In the preterm infant study, the identified microbes are mostly consistent with existing studies and biological understandings. Our approach supports that stressful early life experiences imprint gut microbiome through the regulation of the gut-brain axis.

stat.ME↗

arXiv2019

Unicoder-VL: A Universal Encoder for Vision and Language by Cross-modal Pre-training

We propose Unicoder-VL, a universal encoder that aims to learn joint representations of vision and language in a pre-training manner. Borrow ideas from cross-lingual pre-trained models, such as XLM and Unicoder, both visual and linguistic contents are fed into a multi-layer Transformer for the cross-modal pre-training, where three pre-trained tasks are employed, including Masked Language Modeling (MLM), Masked Object Classification (MOC) and Visual-linguistic Matching (VLM). The first two tasks learn context-aware representations for input tokens based on linguistic and visual contents jointly. The last task tries to predict whether an image and a text describe each other. After pretraining on large-scale image-caption pairs, we transfer Unicoder-VL to caption-based image-text retrieval and visual commonsense reasoning, with just one additional output layer. We achieve state-of-the-art or comparable results on both two tasks and show the powerful ability of the cross-modal pre-training.

cs.CV↗

arXiv2019

Zero-inflated Poisson Factor Model with Application to Microbiome Absolute Abundance Data

Dimension reduction of high-dimensional microbiome data facilitates subsequent analysis such as regression and clustering. Most existing reduction methods cannot fully accommodate the special features of the data such as count-valued and excessive zero reads. We propose a zero-inflated Poisson factor analysis (ZIPFA) model in this article. The model assumes that microbiome absolute abundance data follow zero-inflated Poisson distributions with library size as offset and Poisson rates negatively related to the inflated zero occurrences. The latent parameters of the model form a low-rank matrix consisting of interpretable loadings and low-dimensional scores which can be used for further analyses. We develop an efficient and robust expectation-maximization (EM) algorithm for parameter estimation. We demonstrate the efficacy of the proposed method using comprehensive simulation studies. The application to the Oral Infections, Glucose Intolerance and Insulin Resistance Study (ORIGINS) provides valuable insights into the relation between subgingival microbiome and periodontal disease.

stat.ME↗

arXiv2019

Log-Contrast Regression with Functional Compositional Predictors: Linking Preterm Infant's Gut Microbiome Trajectories to Neurobehavioral Outcome

The neonatal intensive care unit (NICU) experience is known to be one of the most crucial factors that drive preterm infant's neurodevelopmental and health outcomes. It is hypothesized that stressful early life experience of very preterm neonate is imprinting gut microbiome by the regulation of the so-called brain-gut axis, and consequently, certain microbiome markers are predictive of later infant neurodevelopment. To investigate, a preterm infant study was conducted; infant fecal samples were collected during the infants' first month of postnatal age, resulting in functional compositional microbiome data, and neurobehavioral outcomes were measured when infants reached 36-38 weeks of post-menstrual age. To identify potential microbiome markers and estimate how the trajectories of gut microbiome compositions during early postnatal stage impact later neurobehavioral outcomes of the preterm infants, we innovate a sparse log-contrast regression with functional compositional predictors. The functional simplex structure is strictly preserved, and the functional compositional predictors are allowed to have sparse, smoothly varying, and accumulating effects on the outcome through time. Through a pragmatic basis expansion step, the problem boils down to a linearly constrained sparse group regression, for which we develop an efficient algorithm and obtain theoretical performance guarantees. Our approach yields insightful results in the preterm infant study. The identified microbiome markers and the estimated time dynamics of their impact on the neurobehavioral outcome shed light on the linkage between stress accumulation in early postnatal stage and neurodevelopmental process of infants.

stat.ME↗

arXiv2019

DABNet: Depth-wise Asymmetric Bottleneck for Real-time Semantic Segmentation

As a pixel-level prediction task, semantic segmentation needs large computational cost with enormous parameters to obtain high performance. Recently, due to the increasing demand for autonomous systems and robots, it is significant to make a tradeoff between accuracy and inference speed. In this paper, we propose a novel Depthwise Asymmetric Bottleneck (DAB) module to address this dilemma, which efficiently adopts depth-wise asymmetric convolution and dilated convolution to build a bottleneck structure. Based on the DAB module, we design a Depth-wise Asymmetric Bottleneck Network (DABNet) especially for real-time semantic segmentation, which creates sufficient receptive field and densely utilizes the contextual information. Experiments on Cityscapes and CamVid datasets demonstrate that the proposed DABNet achieves a balance between speed and precision. Specifically, without any pretrained model and postprocessing, it achieves 70.1% Mean IoU on the Cityscapes test dataset with only 0.76 million parameters and a speed of 104 FPS on a single GTX 1080Ti card.

cs.CV↗

arXiv2019

Unraveling the Veil of Subspace RIP Through Near-Isometry on Subspaces

Dimensionality reduction is a popular approach to tackle high-dimensional data with low-dimensional nature. Subspace Restricted Isometry Property, a newly-proposed concept, has proved to be a useful tool in analyzing the effect of dimensionality reduction algorithms on subspaces. In this paper, we provide a characterization of subspace Restricted Isometry Property, asserting that matrices which act as a near-isometry on low-dimensional subspaces possess subspace Restricted Isometry Property. This points out a unified approach to discuss subspace Restricted Isometry Property. Its power is further demonstrated by the possibility to prove with this result the subspace RIP for a large variety of random matrices encountered in theory and practice, including subgaussian matrices, partial Fourier matrices, partial Hadamard matrices, partial circulant/Toeplitz matrices, matrices with independent strongly regular rows (for instance, matrices with independent entries having uniformly bounded $4+ε$ moments), and log-concave ensembles. Thus our result could extend the applicability of random projections in subspace-based machine learning algorithms including subspace clustering and allow for the application of some useful random matrices which are easier to implement on hardware or are more efficient to compute.

cs.IT↗

arXiv2019

Theory of Spectral Method for Union of Subspaces-Based Random Geometry Graph

Spectral Method is a commonly used scheme to cluster data points lying close to Union of Subspaces by first constructing a Random Geometry Graph, called Subspace Clustering. This paper establishes a theory to analyze this method. Based on this theory, we demonstrate the efficiency of Subspace Clustering in fairly broad conditions. The insights and analysis techniques developed in this paper might also have implications for other random graph problems. Numerical experiments demonstrate the effectiveness of our theoretical study.

cs.LG↗