arXiv subjects

Yang Yu

Publications and source records attributed to Yang Yu.

At least 433 records · Page 24Linked to original sources

arXiv2019

Cascaded Algorithm-Selection and Hyper-Parameter Optimization with Extreme-Region Upper Confidence Bound Bandit

An automatic machine learning (AutoML) task is to select the best algorithm and its hyper-parameters simultaneously. Previously, the hyper-parameters of all algorithms are joint as a single search space, which is not only huge but also redundant, because many dimensions of hyper-parameters are irrelevant with the selected algorithms. In this paper, we propose a cascaded approach for algorithm selection and hyper-parameter optimization. While a search procedure is employed at the level of hyper-parameter optimization, a bandit strategy runs at the level of algorithm selection to allocate the budget based on the search feedbacks. Since the bandit is required to select the algorithm with the maximum performance, instead of the average performance, we thus propose the extreme-region upper confidence bound (ER-UCB) strategy, which focuses on the extreme region of the underlying feedback distribution. We show theoretically that the ER-UCB has a regret upper bound $O\left(K \ln n\right)$ with independent feedbacks, which is as efficient as the classical UCB bandit. We also conduct experiments on a synthetic problem as well as a set of AutoML tasks. The results verify the effectiveness of the proposed method.

cs.LG↗

arXiv2019

Reinforcement Learning Experience Reuse with Policy Residual Representation

Experience reuse is key to sample-efficient reinforcement learning. One of the critical issues is how the experience is represented and stored. Previously, the experience can be stored in the forms of features, individual models, and the average model, each lying at a different granularity. However, new tasks may require experience across multiple granularities. In this paper, we propose the policy residual representation (PRR) network, which can extract and store multiple levels of experience. PRR network is trained on a set of tasks with a multi-level architecture, where a module in each level corresponds to a subset of the tasks. Therefore, the PRR network represents the experience in a spectrum-like way. When training on a new task, PRR can provide different levels of experience for accelerating the learning. We experiment with the PRR network on a set of grid world navigation tasks, locomotion tasks, and fighting tasks in a video game. The results show that the PRR network leads to better reuse of experience and thus outperforms some state-of-the-art approaches.

cs.LG↗

arXiv2019

Knowledge-augmented Column Networks: Guiding Deep Learning with Advice

Recently, deep models have had considerable success in several tasks, especially with low-level representations. However, effective learning from sparse noisy samples is a major challenge in most deep models, especially in domains with structured representations. Inspired by the proven success of human guided machine learning, we propose Knowledge-augmented Column Networks, a relational deep learning framework that leverages human advice/knowledge to learn better models in presence of sparsity and systematic noise.

cs.LG↗

arXiv2019

Computer-aided Detection of Squamous Carcinoma of the Cervix in Whole Slide Images

Goal: Squamous cell carcinoma of cervix is one of the most prevalent cancer worldwide in females. Traditionally, the most indispensable diagnosis of cervix squamous carcinoma is histopathological assessment which is achieved under microscope by pathologist. However, human evaluation of pathology slide is highly depending on the experience of pathologist, thus big inter- and intra-observer variability exists. Digital pathology, in combination with deep learning provides an opportunity to improve the objectivity and efficiency of histopathologic slide analysis. Methods: In this study, we obtained 800 haematoxylin and eosin stained slides from 300 patients suffered from cervix squamous carcinoma. Based on information from morphological heterogeneity in the tumor and its adjacent area, we established deep learning models using popular convolution neural network architectures (inception-v3, InceptionResnet-v2 and Resnet50). Then random forest was introduced to feature extractions and slide-based classification. Results: The overall performance of our proposed models on slide-based tumor discrimination were outstanding with an AUC scores > 0.94. While, location identifications of lesions in whole slide images were mediocre (FROC scores > 0.52) duo to the extreme complexity of tumor tissues. Conclusion: For the first time, our analysis workflow highlighted a quantitative visual-based slide analysis of cervix squamous carcinoma. Significance: This study demonstrates a pathway to assist pathologist and accelerate the diagnosis of patients by utilizing new computational approaches.

cs.CV↗

arXiv2019

Surface Water Formation on the Natural Surface under Super-saturation: from Local Water Balance to Air Pollutant Deposition

Heterogeneous nucleation and subsequent growth of surface water occur on the natural substrate when the water vapor concentration reached the point of super-saturation. This study focuses on the parameterization of super-saturation on the canopy-air interface by field observations monitoring surface water formation (SWF) such as dew and frost in the evergreen shrub at an urban cite during autumn and winter in 2015-2017. Here we show that both the interfacial and vertical temperature differences ranged from 1 to 3 K and were necessary but not sufficient for super-saturated condensation on the natural surface. Excessive supplies of moisture must exist, continuously contribute to the growth of the condensed water embryos, originate from both the local and the external sources such as evapotranspiration and atmospheric advection driven by the reduced air pressure, cause SWF not only on the ground soil but also on the vegetation canopy at 1-2 m height. The super-saturation ratio is mainly determined by the coefficient of thermophoresis deposition, which approaches to 1. SWF on the natural surface is not only an indicator but also a weak cleaner of air pollution. The downward thermophoresis deposition of fine particle and droplets favors SWF and the scavenging of air pollutants. The removal efficiency of the deposition flux during SWF event for [SO42-+NO3-] is estimated by ~0.3 mmol (per [Ca2+] meq)/m2(per leaf area).

physics.ao-ph↗

arXiv2019

Automatic Calibration of Multiple 3D LiDARs in Urban Environments

Multiple LiDARs have progressively emerged on autonomous vehicles for rendering a wide field of view and dense measurements. However, the lack of precise calibration negatively affects their potential applications in localization and perception systems. In this paper, we propose a novel system that enables automatic multi-LiDAR calibration without any calibration target, prior environmental information, and initial values of the extrinsic parameters. Our approach starts with a hand-eye calibration for automatic initialization by aligning the estimated motions of each sensor. The resulting parameters are then refined with an appearance-based method by minimizing a cost function constructed from point-plane correspondences. Experimental results on simulated and real-world data sets demonstrate the reliability and accuracy of our calibration approach. The proposed approach can calibrate a multi-LiDAR system with the rotation and translation errors less than 0.04 [rad] and 0.1 [m] respectively for a mobile platform.

cs.RO↗

arXiv2019

Many-Body Effect of Mesoscopic Localized States in MoS$_2$ Monolayer

Transition metal dichalcogenide monolayers provide an emerging material system to implement quantum photonics with intrinsic two-dimensional excitons or embedded zero-dimensional localized states. Here we demonstrate the mesoscopic localized states between two- and zero- dimensions, which is a many-body system with electron-electron Coulomb interactions. A fine structure splitting is observed, which is similar to quantum dots. Meanwhile the polarization is changed by the magnetic field, due to the nature of two-dimensional monolayers. Furthermore, a large quadratic diamagnetism with a coefficient of around $100\ \mathrm{μeV/T^2}$ is observed, as a unique consequence of the mesoscopic scale. The many-body effect also results in the emission energy variation and linewidth narrowing in the spectrum, which corresponds well to the theoretical analysis. These unique properties indicate the great potential of mesoscopic localized states in many-body physics and quantum photonics.

cond-mat.mes-hall↗

arXiv2019

A Novel Dual-Lidar Calibration Algorithm Using Planar Surfaces

Multiple lidars are prevalently used on mobile vehicles for rendering a broad view to enhance the performance of localization and perception systems. However, precise calibration of multiple lidars is challenging since the feature correspondences in scan points cannot always provide enough constraints. To address this problem, the existing methods require fixed calibration targets in scenes or rely exclusively on additional sensors. In this paper, we present a novel method that enables automatic lidar calibration without these restrictions. Three linearly independent planar surfaces appearing in surroundings is utilized to find correspondences. Two components are developed to ensure the extrinsic parameters to be found: a closed-form solver for initialization and an optimizer for refinement by minimizing a nonlinear cost function. Simulation and experimental results demonstrate the high accuracy of our calibration approach with the rotation and translation errors smaller than 0.05rad and 0.1m respectively.

cs.CV↗

arXiv2019

Human-Guided Learning of Column Networks: Augmenting Deep Learning with Advice

Recently, deep models have been successfully applied in several applications, especially with low-level representations. However, sparse, noisy samples and structured domains (with multiple objects and interactions) are some of the open challenges in most deep models. Column Networks, a deep architecture, can succinctly capture such domain structure and interactions, but may still be prone to sub-optimal learning from sparse and noisy samples. Inspired by the success of human-advice guided learning in AI, especially in data-scarce domains, we propose Knowledge-augmented Column Networks that leverage human advice/knowledge for better learning with noisy/sparse samples. Our experiments demonstrate that our approach leads to either superior overall performance or faster convergence (i.e., both effective and efficient).

cs.LG↗

arXiv2019

Large g factor in bilayer WS$_2$ flakes

The valley of transition metal dichalcogenides provides an additional platform to manipulate spin due to its unique selection rule. Normally, intralayer optical transitions in magnetic field show a Zeeman splitting with g factor of about $-4$. Here we report remarkable valley Zeeman effect exhibited by splitting of excitonic emission in a bilayer WS$_{2}$, with a value of g factor as large as $-16.5$. The observed large g factor results from the interlayer recombination, as the conduction band and valence band are modified in opposite directions by magnetic field in different layers. The interlayer recombination is due to the defect induced inversion symmetry breaking, which is theoretically not accessible in ideal bilayer WS$_{2}$ with inversion symmetry. Large g factor of interlayer emission offers potential benefits for future optical spin control and detection.

cond-mat.mes-hall↗

arXiv2019

An end-to-end Generative Retrieval Method for Sponsored Search Engine --Decoding Efficiently into a Closed Target Domain

In this paper, we present a generative retrieval method for sponsored search engine, which uses neural machine translation (NMT) to generate keywords directly from query. This method is completely end-to-end, which skips query rewriting and relevance judging phases in traditional retrieval systems. Different from standard machine translation, the target space in the retrieval setting is a constrained closed set, where only committed keywords should be generated. We present a Trie-based pruning technique in beam search to address this problem. The biggest challenge in deploying this method into a real industrial environment is the latency impact of running the decoder. Self-normalized training coupled with Trie-based dynamic pruning dramatically reduces the inference time, yielding a speedup of more than 20 times. We also devise an mixed online-offline serving architecture to reduce the latency and CPU consumption. To encourage the NMT to generate new keywords uncovered by the existing system, training data is carefully selected. This model has been successfully applied in Baidu's commercial search engine as a supplementary retrieval branch, which has brought a remarkable revenue improvement of more than 10 percents.

cs.IR↗

arXiv2019

Tuning the carrier tunneling in a single quantum dot with a magnetic field in Faraday geometry

We report on an increase in the carrier tunneling time in a single quantum dot (QD) with a magnetic field in Faraday geometry using photocurrent spectroscopy. A nearly 60\% increase in hole tunneling time is observed with an applied magnetic field equal to 9 T. For a truncated pyramid QD, hole tunnels out faster at the lateral edge of the QD due to the reduced barrier height. The magnetic field in Faraday geometry shrinks the hole wave function at the center of QD plane, which weakens the tunneling at lateral edge and increases the average tunneling time. This mechanism also works for electron but the effect is smaller. The electron wave function is more localized at the center of the QD due to the uniform confining potential, therefore the relatively weak shrinkage caused by the magnetic field does not reduce the tunneling rate significantly.

cond-mat.mes-hall↗

arXiv2019

Enhanced Strong Interaction between Nanocavities and p-shell Excitons Beyond the Dipole Approximation

Large coupling strengths in exciton-photon interactions are important for quantum photonic network, while strong cavity-quantum-dot interactions have been focused on s-shell excitons with small coupling strengths. Here we demonstrate strong interactions between cavities and p-shell excitons with a great enhancement by the \textit{in situ} wave-function control. The p-shell excitons are demonstrated with much larger wave-function extents and nonlocal interactions beyond the dipole approximation. Then the interaction is tuned from the nonlocal to local regime by the wave-function shrinking, during which the enhancement is obtained. A large coupling strength of $210\ μ\mathrm{eV}$ has been achieved, indicating the great potential of p-shell excitons for coherent information exchange. Furthermore, we propose a distributed delay model to quantitatively explain the coupling strength variation, revealing the intertwining of excitons and photons beyond the dipole approximation.

physics.optics↗

arXiv2019

Minimization of a Class of Rare Event Probabilities and Buffer Probabilities of Exceedance

We consider the problem of choosing design parameters to minimize the probability of an undesired rare event that is described through the average of $n$ iid random variables. Since the probability of interest for near optimal design parameters is very small, one needs to develop suitable accelerated Monte-Carlo methods for estimating the objective function of interest. One of the challenges in the study is that simulating from exponential twists of the laws of the summands may be computationally demanding since these transformed laws may be non-standard and intractable. We consider a setting where the summands are given as a nonlinear functional of random variables that are more tractable for importance sampling in that the exponential twists of their distributions take a simpler form (than that for the original summands). We also study the closely related problem of estimating buffered probability of exceedance and provide the first rigorous results that relate the asymptotics of buffered probability and that of the ordinary probability under a large deviation scaling. The analogous minimization problem for buffered probability, under conditions, can be formulated as a convex optimization problem which makes it more tractable than the original optimization problem. We show that, under conditions, changes of measures that are asymptotically efficient (under the large deviation scaling) for estimating ordinary probability are also asymptotically efficient for estimating the buffered probability of exceedance. We embed the constructed importance sampling scheme in suitable gradient descent/ascent algorithms for solving the optimization problems of interest. Implementation of schemes for some examples is illustrated through computational experiments.

math.OC↗

arXiv2019

PointIT: A Fast Tracking Framework Based on 3D Instance Segmentation

Recently most popular tracking frameworks focus on 2D image sequences. They seldom track the 3D object in point clouds. In this paper, we propose PointIT, a fast, simple tracking method based on 3D on-road instance segmentation. Firstly, we transform 3D LiDAR data into the spherical image with the size of 64 x 512 x 4 and feed it into instance segment model to get the predicted instance mask for each class. Then we use MobileNet as our primary encoder instead of the original ResNet to reduce the computational complexity. Finally, we extend the Sort algorithm with this instance framework to realize tracking in the 3D LiDAR point cloud data. The model is trained on the spherical images dataset with the corresponding instance label masks which are provided by KITTI 3D Object Track dataset. According to the experiment results, our network can achieve on Average Precision (AP) of 0.617 and the performance of multi-tracking task has also been improved.

cs.CV↗

arXiv2019

Tuplemax Loss for Language Identification

In many scenarios of a language identification task, the user will specify a small set of languages which he/she can speak instead of a large set of all possible languages. We want to model such prior knowledge into the way we train our neural networks, by replacing the commonly used softmax loss function with a novel loss function named tuplemax loss. As a matter of fact, a typical language identification system launched in North America has about 95% users who could speak no more than two languages. Using the tuplemax loss, our system achieved a 2.33% error rate, which is a relative 39.4% improvement over the 3.85% error rate of standard softmax loss method.

eess.AS↗

arXiv2019

Giant Photocurrent Enhancement by Coulomb Interaction in a Single Quantum Dot for Energy Harvesting

Understanding the carrier excitation and transport processes at the single-charge level plays a key role in quantum-dot-based solar cells and photodetectors. Here, we report on Coulomb-induced giant photocurrent enhancement of positive charged trions (\emph{X$^+$}) in a single self-assembled InAs/GaAs quantum dot embedded in an \emph{n-i-}Schottky device by high-resolution photocurrent (PC) spectroscopy. The Coulomb repulsion between the two holes in the \emph{X$^+$} increases the tunneling rate of the hole, and the remaining hole can be reused as the initial state to regenerate \emph{X$^+$} again. This process brings the PC amplitude of \emph{X$^+$} up to 30 times larger than that of the neutral exciton. The analysis of the hole tunneling time gives the equivalent change of hole tunnel barriers caused by Coulomb interaction between two holes with a value of 8.05 meV during the tunneling process. Our work brings a fundamental understanding of energy conversion for solar cells in nanoscale to improve internal quantum efficiency for energy harvesting.

cond-mat.mes-hall↗

arXiv2019

On Reinforcement Learning for Full-length Game of StarCraft

StarCraft II poses a grand challenge for reinforcement learning. The main difficulties of it include huge state and action space and a long-time horizon. In this paper, we investigate a hierarchical reinforcement learning approach for StarCraft II. The hierarchy involves two levels of abstraction. One is the macro-action automatically extracted from expert's trajectories, which reduces the action space in an order of magnitude yet remains effective. The other is a two-layer hierarchical architecture which is modular and easy to scale, enabling a curriculum transferring from simpler tasks to more complex tasks. The reinforcement training algorithm for this architecture is also investigated. On a 64x64 map and using restrictive units, we achieve a winning rate of more than 99\% against the difficulty level-1 built-in AI. Through the curriculum transfer learning algorithm and a mixture of combat model, we can achieve over 93\% winning rate of Protoss against the most difficult non-cheating built-in AI (level-7) of Terran, training within two days using a single machine with only 48 CPU cores and 8 K40 GPUs. It also shows strong generalization performance, when tested against never seen opponents including cheating levels built-in AI and all levels of Zerg and Protoss built-in AI. We hope this study could shed some light on the future research of large-scale reinforcement learning.

cs.LG↗