arXiv subjects

Bo Liu

Publications and source records attributed to Bo Liu.

At least 613 records · Page 34Linked to original sources

arXiv2019

Anonymized BERT: An Augmentation Approach to the Gendered Pronoun Resolution Challenge

We present our 7th place solution to the Gendered Pronoun Resolution challenge, which uses BERT without fine-tuning and a novel augmentation strategy designed for contextual embedding token-level tasks. Our method anonymizes the referent by replacing candidate names with a set of common placeholder names. Besides the usual benefits of effectively increasing training data size, this approach diversifies idiosyncratic information embedded in names. Using same set of common first names can also help the model recognize names better, shorten token length, and remove gender and regional biases associated with names. The system scored 0.1947 log loss in stage 2, where the augmentation contributed to an improvements of 0.04. Post-competition analysis shows that, when using different embedding layers, the system scores 0.1799 which would be third place.

cs.CL↗

arXiv2019

Non-uniqueness phenomenon of object representation in modelling IT cortex by deep convolutional neural network (DCNN)

Recently DCNN (Deep Convolutional Neural Network) has been advocated as a general and promising modelling approach for neural object representation in primate inferotemporal cortex. In this work, we show that some inherent non-uniqueness problem exists in the DCNN-based modelling of image object representations. This non-uniqueness phenomenon reveals to some extent the theoretical limitation of this general modelling approach, and invites due attention to be taken in practice.

q-bio.NC↗

arXiv2019

Enhancement of intrinsic magnetic damping in defect-free epitaxial Fe3O4 thin films

We have investigated the magnetic damping of precessional spin dynamics in defect-controlled epitaxial grown Fe$_3$O$_4$(111)/Yttria-stabilized Zirconia (YSZ) nanoscale films by all-optical pump-probe measurements. The intrinsic damping constant of the defect-free Fe$_3$O$_4$ film is found to be strikingly larger than that of the as-grown Fe$_3$O$_4$ film with structural defects. We demonstrate that the population of the first-order perpendicular standing spin wave (PSSW) mode, which is exclusively observed in the defect-free film under sufficiently high external magnetic fields, leads to the enhancement of the magnetic damping of the uniform precession (Kittel) mode. We propose a physical picture in which the PSSW mode acts as an additional channel for the extra energy dissipation of the Kittel mode. The energy transfer from Kittel mode to PSSW mode increases as in-plane magnetization precession becomes more uniform, resulting in the unique intrinsic magnetic damping enhancement in the defect-free Fe$_3$O$_4$ film.

cond-mat.mtrl-sci↗

arXiv2019

Optimal Control of Complex Systems through Variational Inference with a Discrete Event Decision Process

Complex social systems are composed of interconnected individuals whose interactions result in group behaviors. Optimal control of a real-world complex system has many applications, including road traffic management, epidemic prevention, and information dissemination. However, such real-world complex system control is difficult to achieve because of high-dimensional and non-linear system dynamics, and the exploding state and action spaces for the decision maker. Prior methods can be divided into two categories: simulation-based and analytical approaches. Existing simulation approaches have high-variance in Monte Carlo integration, and the analytical approaches suffer from modeling inaccuracy. We adopted simulation modeling in specifying the complex dynamics of a complex system, and developed analytical solutions for searching optimal strategies in a complex network with high-dimensional state-action space. To capture the complex system dynamics, we formulate the complex social network decision making problem as a discrete event decision process. To address the curse of dimensionality and search in high-dimensional state action spaces in complex systems, we reduce control of a complex system to variational inference and parameter learning, introduce Bethe entropy approximation, and develop an expectation propagation algorithm. Our proposed algorithm leads to higher system expected rewards, faster convergence, and lower variance of value function in a real-world transportation scenario than state-of-the-art analytical and sampling approaches.

eess.SY↗

arXiv2019

Feature Space Transfer for Data Augmentation

The problem of data augmentation in feature space is considered. A new architecture, denoted the FeATure TransfEr Network (FATTEN), is proposed for the modeling of feature trajectories induced by variations of object pose. This architecture exploits a parametrization of the pose manifold in terms of pose and appearance. This leads to a deep encoder/decoder network architecture, where the encoder factors into an appearance and a pose predictor. Unlike previous attempts at trajectory transfer, FATTEN can be efficiently trained end-to-end, with no need to train separate feature transfer functions. This is realized by supplying the decoder with information about a target pose and the use of a multi-task loss that penalizes category- and pose-mismatches. In result, FATTEN discourages discontinuous or non-smooth trajectories that fail to capture the structure of the pose manifold, and generalizes well on object recognition tasks involving large pose variation. Experimental results on the artificial ModelNet database show that it can successfully learn to map source features to target features of a desired pose, while preserving class identity. Most notably, by using feature space transfer for data augmentation (w.r.t. pose and depth) on SUN-RGBD objects, we demonstrate considerable performance improvements on one/few-shot object recognition in a transfer learning setup, compared to current state-of-the-art methods.

cs.CV↗

arXiv2019

Predicting pregnancy using large-scale data from a women's health tracking mobile application

Predicting pregnancy has been a fundamental problem in women's health for more than 50 years. Previous datasets have been collected via carefully curated medical studies, but the recent growth of women's health tracking mobile apps offers potential for reaching a much broader population. However, the feasibility of predicting pregnancy from mobile health tracking data is unclear. Here we develop four models -- a logistic regression model, and 3 LSTM models -- to predict a woman's probability of becoming pregnant using data from a women's health tracking app, Clue by BioWink GmbH. Evaluating our models on a dataset of 79 million logs from 65,276 women with ground truth pregnancy test data, we show that our predicted pregnancy probabilities meaningfully stratify women: women in the top 10% of predicted probabilities have a 89% chance of becoming pregnant over 6 menstrual cycles, as compared to a 27% chance for women in the bottom 10%. We develop a technique for extracting interpretable time trends from our deep learning models, and show these trends are consistent with previous fertility research. Our findings illustrate the potential that women's health tracking data offers for predicting pregnancy on a broader population; we conclude by discussing the steps needed to fulfill this potential.

stat.AP↗

arXiv2019

Oblique droplet impact onto a deep liquid pool

The oblique impact of a liquid droplet onto a deep liquid pool is studied numerically with the adaptive volume-of-fluid solver Basilisk. The splashing threshold, cavity formation, cavity evolution and the maximum cavity dimensions are quantified as a function of the Weber number and the impact angle. We compare the numerical results with recent experimental work by Gielen et al. [Phys. Rev. Fluids 2, 083602 (2017)]. Similarly to the experimental results, three different impact regimes are observed: deposition of the droplet onto the pool, single-sided splashing in the direction of the impact and splashing in all directions. We show good qualitative and quantitative agreement of the splashing behaviour and cavity formation between the simulations and the experiments. Furthermore, the simulations provide a three dimensional view of the impact phenomenon, give access to velocity and pressure fields, and allow to explore impact parameters that are hard to achieve experimentally.

physics.flu-dyn↗

arXiv2019

Constraining the anomalous Higgs boson coupling in $H$+$γ$ production

Higgs boson production in association with a photon ($H$+$γ$) offers a promising channel to test the Higgs boson to photon coupling at various energy scales. Its potential sensitivity to anomalous couplings of the Higgs boson has not been explored with the proton-proton collision data. In this paper, we reinterpret the latest ATLAS $H$+$γ$ resonance search results within the Standard Model effective field theory (EFT) framework, using 36.1 fb$^{-1}$ of proton-proton collision data recorded with the ATLAS detector at $\sqrt{s}=13$ TeV. Constraints on the Wilson coefficients of dimension-six EFT operators related to the Higgs boson to photon coupling are provided for the first time in the $H$+$γ$ final state at the LHC.

hep-ph↗

arXiv2019

Precision Higgs Physics at CEPC

The discovery of the Higgs boson with its mass around 125 GeV by the ATLAS and CMS Collaborations marked the beginning of a new era in high energy physics. The Higgs boson will be the subject of extensive studies of the ongoing LHC program. At the same time, lepton collider based Higgs factories have been proposed as a possible next step beyond the LHC, with its main goal to precisely measure the properties of the Higgs boson and probe potential new physics associated with the Higgs boson. The Circular Electron Positron Collider~(CEPC) is one of such proposed Higgs factories. The CEPC is an $e^+e^-$ circular collider proposed by and to be hosted in China. Located in a tunnel of approximately 100~km in circumference, it will operate at a center-of-mass energy of 240~GeV as the Higgs factory. In this paper, we present the first estimates on the precision of the Higgs boson property measurements achievable at the CEPC and discuss implications of these measurements.

hep-ex↗

arXiv2019

SDRL: Interpretable and Data-efficient Deep Reinforcement Learning Leveraging Symbolic Planning

Deep reinforcement learning (DRL) has gained great success by learning directly from high-dimensional sensory inputs, yet is notorious for the lack of interpretability. Interpretability of the subtasks is critical in hierarchical decision-making as it increases the transparency of black-box-style DRL approach and helps the RL practitioners to understand the high-level behavior of the system better. In this paper, we introduce symbolic planning into DRL and propose a framework of Symbolic Deep Reinforcement Learning (SDRL) that can handle both high-dimensional sensory inputs and symbolic planning. The task-level interpretability is enabled by relating symbolic actions to options.This framework features a planner -- controller -- meta-controller architecture, which takes charge of subtask scheduling, data-driven subtask learning, and subtask evaluation, respectively. The three components cross-fertilize each other and eventually converge to an optimal symbolic plan along with the learned subtasks, bringing together the advantages of long-term planning capability with symbolic knowledge and end-to-end reinforcement learning directly from a high-dimensional sensory input. Experimental results validate the interpretability of subtasks, along with improved data efficiency compared with state-of-the-art approaches.

cs.AI↗

arXiv2018

A Container-based DoS Attack-Resilient Control Framework for Real-Time UAV Systems

The Unmanned aerial vehicles (UAVs) sector is fast-expanding. Protection of real-time UAV applications against malicious attacks has become an urgent problem that needs to be solved. Denial-of-service (DoS) attack aims to exhaust system resources and cause important tasks to miss deadlines. DoS attack may be one of the common problems of UAV systems, due to its simple implementation. In this paper, we present a software framework that offers DoS attack-resilient control for real-time UAV systems using containers: ContainerDrone. The framework provides defense mechanisms for three critical system resources: CPU, memory, and communication channel. We restrict attacker's access to CPU core set and utilization. Memory bandwidth throttling limits attacker's memory usage. By simulating sensors and drivers in the container, a security monitor constantly checks DoS attacks over communication channels. Upon the detection of a security rule violation, the framework switches to the safety controller to mitigate the attack. We implemented a prototype quadcopter with commercially off-the-shelf (COTS) hardware and open-source software. Our experimental results demonstrated the effectiveness of the proposed framework defending against various DoS attacks.

cs.CR↗

arXiv2018

Viconmavlink: A software tool for indoor positioning using a motion capture system

Motion capture is a widely-used technology in robotics research thanks to its precise posi tional measurements with real-time performance. This paper presents ViconMAVLink, a cross-platform open-source software tool that provides indoor positioning services to networked robots. ViconMAVLink converts Vicon motion capture data into proper pose and motion data formats and send localization information to robots using the MAVLink protocol. The software is a convenient tool for mobile robotics researchers to conduct experiments in a controlled indoor environment.

cs.RO↗

arXiv2018

Constrained-size Tensorflow Models for YouTube-8M Video Understanding Challenge

This paper presents our 7th place solution to the second YouTube-8M video understanding competition which challenges participates to build a constrained-size model to classify millions of YouTube videos into thousands of classes. Our final model consists of four single models aggregated into one tensorflow graph. For each single model, we use the same network architecture as in the winning solution of the first YouTube-8M video understanding competition, namely Gated NetVLAD. We train the single models separately in tensorflow's default float32 precision, then replace weights with float16 precision and ensemble them in the evaluation and inference stages., achieving 48.5% compression rate without loss of precision. Our best model achieved 88.324% GAP on private leaderboard. The code is publicly available at https://github.com/boliu61/youtube-8m

cs.CV↗

arXiv2018

QUOTA: The Quantile Option Architecture for Reinforcement Learning

In this paper, we propose the Quantile Option Architecture (QUOTA) for exploration based on recent advances in distributional reinforcement learning (RL). In QUOTA, decision making is based on quantiles of a value distribution, not only the mean. QUOTA provides a new dimension for exploration via making use of both optimism and pessimism of a value distribution. We demonstrate the performance advantage of QUOTA in both challenging video games and physical robot simulators.

cs.LG↗

arXiv2018

Dantzig Selector with an Approximately Optimal Denoising Matrix and its Application to Reinforcement Learning

Dantzig Selector (DS) is widely used in compressed sensing and sparse learning for feature selection and sparse signal recovery. Since the DS formulation is essentially a linear programming optimization, many existing linear programming solvers can be simply applied for scaling up. The DS formulation can be explained as a basis pursuit denoising problem, wherein the data matrix (or measurement matrix) is employed as the denoising matrix to eliminate the observation noise. However, we notice that the data matrix may not be the optimal denoising matrix, as shown by a simple counter-example. This motivates us to pursue a better denoising matrix for defining a general DS formulation. We first define the optimal denoising matrix through a minimax optimization, which turns out to be an NPhard problem. To make the problem computationally tractable, we propose a novel algorithm, termed as Optimal Denoising Dantzig Selector (ODDS), to approximately estimate the optimal denoising matrix. Empirical experiments validate the proposed method. Finally, a novel sparse reinforcement learning algorithm is formulated by extending the proposed ODDS algorithm to temporal difference learning, and empirical experimental results demonstrate to outperform the conventional vanilla DS-TD algorithm.

cs.LG↗

arXiv2018

A Block Coordinate Ascent Algorithm for Mean-Variance Optimization

Risk management in dynamic decision problems is a primary concern in many fields, including financial investment, autonomous driving, and healthcare. The mean-variance function is one of the most widely used objective functions in risk management due to its simplicity and interpretability. Existing algorithms for mean-variance optimization are based on multi-time-scale stochastic approximation, whose learning rate schedules are often hard to tune, and have only asymptotic convergence proof. In this paper, we develop a model-free policy search framework for mean-variance optimization with finite-sample error bound analysis (to local optima). Our starting point is a reformulation of the original mean-variance function with its Fenchel dual, from which we propose a stochastic block coordinate ascent policy search algorithm. Both the asymptotic convergence guarantee of the last iteration's solution and the convergence rate of the randomly picked solution are provided, and their applicability is demonstrated on several benchmark domains.

cs.LG↗

arXiv2018

Parameter Synthesis Problems for Parametric Timed Automata

We consider the parameter synthesis problem of parametric timed automata (PTAs). The problem is, given a PTA and a property, to compute the set of valuations of the parameters under which the resulting timed automaton satisfies the property. Such a set of parameter valuations is called a feasible region for the PTA and the property. The problem is known undecidable in general. This paper, however, presents our study on some decidable sub-classes of PTAs and proposes efficient parameter synthesis algorithms for them.

eess.SY↗

arXiv2018

Novel nonlinear electrodynamics black hole and related phenomena in the extended thermodynamics

In this paper, we present the analytic solution for Anti-de Sitter-Nonlinear electrodynamics (AdS-NED) black hole in four dimensions. We study tunneling of vector particles from the AdS-NED black hole to find Hawking temperature as well as its thermodynamical properties in usual and extended phase space. Then we explore the properties of pressure-volume criticality by analyzing the phase diagram and its free energy. Finally, we investigate the Joule-Thomson coefficient, in which the inversion curve of warming and cooling are studied, and the ratio between the inverse temperature and critical temperature are computed in the black hole solution.

gr-qc↗