arXiv subjects

Tianyi Zhang

Publications and source records attributed to Tianyi Zhang.

At least 361 records · Page 20Linked to original sources

arXiv2021

A One-step Approach to Covariate Shift Adaptation

A default assumption in many machine learning scenarios is that the training and test samples are drawn from the same probability distribution. However, such an assumption is often violated in the real world due to non-stationarity of the environment or bias in sample selection. In this work, we consider a prevalent setting called covariate shift, where the input distribution differs between the training and test stages while the conditional distribution of the output given the input remains unchanged. Most of the existing methods for covariate shift adaptation are two-step approaches, which first calculate the importance weights and then conduct importance-weighted empirical risk minimization. In this paper, we propose a novel one-step approach that jointly learns the predictive model and the associated weights in one optimization by minimizing an upper bound of the test risk. We theoretically analyze the proposed method and provide a generalization error bound. We also empirically demonstrate the effectiveness of the proposed method.

cs.LG↗

arXiv2021

On the Inductive Bias of Masked Language Modeling: From Statistical to Syntactic Dependencies

We study how masking and predicting tokens in an unsupervised fashion can give rise to linguistic structures and downstream performance gains. Recent theories have suggested that pretrained language models acquire useful inductive biases through masks that implicitly act as cloze reductions for downstream tasks. While appealing, we show that the success of the random masking strategy used in practice cannot be explained by such cloze-like masks alone. We construct cloze-like masks using task-specific lexicons for three different classification datasets and show that the majority of pretrained performance gains come from generic masks that are not associated with the lexicon. To explain the empirical success of these generic masks, we demonstrate a correspondence between the Masked Language Model (MLM) objective and existing methods for learning statistical dependencies in graphical models. Using this, we derive a method for extracting these learned statistical dependencies in MLMs and show that these dependencies encode useful inductive biases in the form of syntactic structures. In an unsupervised parsing evaluation, simply forming a minimum spanning tree on the implied statistical dependence structure outperforms a classic method for unsupervised parsing (58.74 vs. 55.91 UUAS).

cs.CL↗

arXiv2021

Fully Dynamic Maximal Independent Set in Expected Poly-Log Update Time

In the fully dynamic maximal independent set (MIS) problem our goal is to maintain an MIS in a given graph $G$ while edges are inserted and deleted from the graph. The first non-trivial algorithm for this problem was presented by Assadi, Onak, Schieber, and Solomon [STOC 2018] who obtained a deterministic fully dynamic MIS with $O(m^{3/4})$ update time. Later, this was independently improved by Du and Zhang and by Gupta and Khan [arXiv 2018] to $\tilde{O}(m^{2/3})$ update time. Du and Zhang [arXiv 2018] also presented a randomized algorithm against an oblivious adversary with $\tilde{O}(\sqrt{m})$ update time. The current state of art is by Assadi, Onak, Schieber, and Solomon [SODA 2019] who obtained randomized algorithms against oblivious adversary with $\tilde{O}(\sqrt{n})$ and $\tilde{O}(m^{1/3})$ update times. In this paper, we propose a dynamic randomized algorithm against oblivious adversary with expected worst-case update time of $O(\log^4n)$. As a direct corollary, one can apply the black-box reduction from a recent work by Bernstein, Forster, and Henzinger [SODA 2019] to achieve $O(\log^6n)$ worst-case update time with high probability. This is the first dynamic MIS algorithm with very fast update time of poly-log.

cs.DS↗

arXiv2021

Design Paradigms Based on Spring Agonists for Underactuated Robot Hands: Concepts and Application

In this paper, we focus on a rarely used paradigm in the design of underactuated robot hands: the use of springs as agonists and tendons as antagonists. We formalize this approach in a design matrix also considering its interplay with the underactuation method used (one tendon for multiple joints vs. multiple tendons on one motor shaft). We then show how different cells in this design matrix can be combined in order to facilitate the implementation of desired postural synergies with a single motor. Furthermore, we show that when agonist and antagonist tendons are combined on the same motor shaft, the resulting spring force cancellation can be leveraged to produce multiple desirable behaviors, which we demonstrate in a physical prototype.

cs.RO↗

arXiv2021

RFI Mitigation for One-bit UWB Radar Systems

Radio frequency interference (RFI) mitigation is critical to the proper operation of ultra-wideband (UWB) radar systems since RFI can severely degrade the radar imaging capability and target detection performance. In this paper, we address the RFI mitigation problem for one-bit UWB radar systems. A one-bit UWB system obtains its signed measurements via a low-cost and high rate sampling scheme, referred to as the Continuous Time Binary Value (CTBV) technology. This sampling strategy compares the signal to a known threshold varying with slow-time and therefore can be used to achieve a rather high sampling rate and quantization resolution with rather simple and affordable hardware. This paper establishes a proper data model for the RFI sources and proposes a novel RFI mitigation method for the one-bit UWB radar system that uses the CTBV sampling technique. Specifically, we first model the RFI sources as a sum of sinusoids with frequencies fixed during the coherent processing interval (CPI) and we exploit the sparsity of the RFI spectrum. We extend a majorization-minimization based 1bRELAX algorithm, referred to as 1bMMRELAX, to estimate the RFI source parameters from the signed measurements obtained by using the CTBV sampling strategy. We also devise a new fast frequency initialization method based on the Alternating Direction Method of Multipliers (ADMM) methodology for the extended 1bMMRELAX algorithm to significantly improve its computational efficiency. Moreover, an ADMM-based sparse method is introduced to recover the desired radar echoes using the estimated RFI parameters. Both simulated and experimental results are presented to demonstrate that our proposed algorithm outperforms the existing digital integration method, especially for severe RFI cases.

eess.SP↗

arXiv2021

Sinusoidal Parameter Estimation from Signed Measurements via Majorization-Minimization Based RELAX

We consider the problem of sinusoidal parameter estimation using signed observations obtained via one-bit sampling with fixed as well as time-varying thresholds. In a previous paper, a relaxation-based algorithm, referred to as 1bRELAX, has been proposed to iteratively maximize the likelihood function. However, the exhaustive search procedure used in each iteration of 1bRELAX is time-consuming. In this paper, we present a majorization-minimization (MM) based 1bRELAX algorithm, referred to as 1bMMRELAX, to enhance the computational efficiency of 1bRELAX. Using the MM technique, 1bMMRELAX maximizes the likelihood function iteratively using simple FFT operations instead of the more computationally intensive search used by 1bRELAX. Both simulated and experimental results are presented to show that 1bMMRELAX can significantly reduce the computational cost of 1bRELAX while maintaining its excellent estimation accuracy.

eess.SP↗

arXiv2021

Joint RFI Mitigation and Radar Echo Recovery for One-Bit UWB Radar

Radio frequency interference (RFI) mitigation and radar echo recovery are critically important for the proper functioning of ultra-wideband (UWB) radar systems using one-bit sampling techniques. We recently introduced a technique for one-bit UWB radar, which first uses a majorization-minimization method for RFI parameter estimation followed by a sparse method for radar echo recovery. However, this technique suffers from high computational complexity due to the need to estimate the parameters of each RFI source separately and iteratively. In this paper, we present a computationally efficient joint RFI mitigation and radar echo recovery framework to greatly reduce the computational cost. Specifically, we exploit the sparsity of RFI in the fast-frequency domain and the sparsity of radar echoes in the fast-time domain to design a one-bit weighted SPICE (SParse Iterative Covariance-based Estimation) based framework for the joint RFI mitigation and radar echo recovery of one-bit UWB radar. Both simulated and experimental results are presented to show that the proposed one-bit weighted SPICE framework can not only reduce the computational cost but also outperform the existing approach for decoupled RFI mitigation and radar echo recovery of one-bit UWB radar.

eess.SP↗

arXiv2021

Revisiting Few-sample BERT Fine-tuning

This paper is a study of fine-tuning of BERT contextual representations, with focus on commonly observed instabilities in few-sample scenarios. We identify several factors that cause this instability: the common use of a non-standard optimization method with biased gradient estimation; the limited applicability of significant parts of the BERT network for down-stream tasks; and the prevalent practice of using a pre-determined, and small number of training iterations. We empirically test the impact of these factors, and identify alternative practices that resolve the commonly observed instability of the process. In light of these observations, we re-visit recently proposed methods to improve few-sample fine-tuning with BERT and re-evaluate their effectiveness. Generally, we observe the impact of these methods diminishes significantly with our modified process.

cs.CL↗

arXiv2021

Learning to Stop with Surprisingly Few Samples

We consider a discounted infinite horizon optimal stopping problem. If the underlying distribution is known a priori, the solution of this problem is obtained via dynamic programming (DP) and is given by a well known threshold rule. When information on this distribution is lacking, a natural (though naive) approach is "explore-then-exploit," whereby the unknown distribution or its parameters are estimated over an initial exploration phase, and this estimate is then used in the DP to determine actions over the residual exploitation phase. We show: (i) with proper tuning, this approach leads to performance comparable to the full information DP solution; and (ii) despite common wisdom on the sensitivity of such "plug in" approaches in DP due to propagation of estimation errors, a surprisingly "short" (logarithmic in the horizon) exploration horizon suffices to obtain said performance. In cases where the underlying distribution is heavy-tailed, these observations are even more pronounced: a ${\it single \, sample}$ exploration phase suffices.

cs.LG↗

arXiv2021

Second- and third-order optical susceptibilities in bidimensional semiconductors near excitons states

Semiconducting Transition Metal Dichalcogenides (TMDs) have significant nonlinear optical effects. In this work we have used second-harmonic generation (SHG) and the four-wave mixing (FWM) spectroscopy in resonance with the excitons in MoS2, MoSe2, and WS2 monolayers to characterize the nonlinear optical properties of these materials. We show that trions and excitons are responsible for enhancing the nonlinear optical response, and determine the exciton and trion energies by comparing with the photoluminescence spectra. Moreover, we extract the second and third order optical sheet susceptibility near exciton energies and compare with values found in the literature. We also demonstrate the ability to generate different nonlinear effects in a wide spectral range in the visible region for monolayer MoS2, opening the possibility of using two-dimensional materials for nonlinear optoelectronic and photonic applications.

cond-mat.mes-hall↗

arXiv2020

Identifying Mislabeled Data using the Area Under the Margin Ranking

Not all data in a typical training set help with generalization; some samples can be overly ambiguous or outrightly mislabeled. This paper introduces a new method to identify such samples and mitigate their impact when training neural networks. At the heart of our algorithm is the Area Under the Margin (AUM) statistic, which exploits differences in the training dynamics of clean and mislabeled samples. A simple procedure - adding an extra class populated with purposefully mislabeled threshold samples - learns a AUM upper bound that isolates mislabeled data. This approach consistently improves upon prior work on synthetic and real-world datasets. On the WebVision50 classification task our method removes 17% of training data, yielding a 1.6% (absolute) improvement in test error. On CIFAR100 removing 13% of the data leads to a 1.2% drop in error.

cs.LG↗

arXiv2020

Physical Wireless Resource Virtualization for Software-Defined Whole-Stack Slicing

Radio access network (RAN) virtualization is gaining more and more ground and expected to re-architect the next-generation cellular networks. Existing RAN virtualization studies and solutions have mostly focused on sharing communication capacity and tend to require the use of the same PHY and MAC layers across network slices. This approach has not considered the scenarios where different slices require different PHY and MAC layers, for instance, for radically different services and for whole-stack research in wireless living labs where novel PHY and MAC layers need to be deployed concurrently with existing ones on the same physical infrastructure. To enable whole-stack slicing where different PHY and MAC layers may be deployed in different slices, we develop PV-RAN, the first open-source virtual RAN platform that enables the sharing of the same SDR physical resources across multiple slices. Through API Remoting, PV-RAN enables running paravirtualized instances of OpenAirInterface (OAI) at different slices without requiring modifying OAI source code. PV-RAN effectively leverages the inter-domain communication mechanisms of Xen to transport time-sensitive I/Q samples via shared memory, making the virtualization overhead in communication almost negligible. We conduct detailed performance benchmarking of PV-RAN and demonstrate its low overhead and high efficiency. We also integrate PV-RAN with the CyNet wireless living lab for smart agriculture and transportation.

cs.NI↗

arXiv2020

Generative Adversarial Network based Heuristics for Sampling-based Path Planning

Sampling-based path planning is a popular methodology for robot path planning. With a uniform sampling strategy to explore the state space, a feasible path can be found without the complex geometric modeling of the configuration space. However, the quality of initial solution is not guaranteed and the convergence speed to the optimal solution is slow. In this paper, we present a novel image-based path planning algorithm to overcome these limitations. Specifically, a generative adversarial network (GAN) is designed to take the environment map (denoted as RGB image) as the input without other preprocessing works. The output is also an RGB image where the promising region (where a feasible path probably exists) is segmented. This promising region is utilized as a heuristic to achieve nonuniform sampling for the path planner. We conduct a number of simulation experiments to validate the effectiveness of the proposed method, and the results demonstrate that our method performs much better in terms of the quality of initial solution and the convergence speed to the optimal solution. Furthermore, apart from the environments similar to the training set, our method also works well on the environments which are very different from the training set.

cs.RO↗

arXiv2020

Exempla Gratis (E.G.): Code Examples for Free

Modern software engineering often involves using many existing APIs, both open source and, in industrial coding environments, proprietary. Programmers reference documentation and code search tools to remind themselves of proper common usage patterns of APIs. However, high-quality API usage examples are computationally expensive to curate and maintain, and API usage examples retrieved from company-wide code search can be tedious to review. We present a tool, EG, that mines codebases and shows the common, idiomatic usage examples for API methods. EG was integrated into Facebook's internal code search tool for the Hack language and evaluated on open-source GitHub projects written in Python. EG was also compared against code search results and hand-written examples from a popular programming website called ProgramCreek. Compared with these two baselines, examples generated by EG are more succinct and representative with less extraneous statements. In addition, a survey with Facebook developers shows that EG examples are preferred in 97 percent of cases.

cs.SE↗

arXiv2020

Can Steering Wheel Detect Your Driving Fatigue?

Automated Driving System (ADS) has attracted increasing attention from both industrial and academic communities due to its potential for increasing the safety, mobility and efficiency of existing transportation systems. The state-of-the-art ADS follows the human-in-the-loop (HITL) design, where the driver's anomalous behaviour is closely monitored by the system. Though many approaches have been proposed for detecting driver fatigue, they largely depend on vehicle driving parameters and facial features, which lacks reliability. Approaches using physiological based sensors (e.g., electroencephalogram or electrocardiogram) are either too clumsy to wear or impractical to install. In this paper, we propose a novel driver fatigue detection method by embedding surface electromyography (sEMG) sensors on a steering wheel. Compared with the existing methods, our approach is able to collect bio-signals in a non-intrusive way and detect driver fatigue at an earlier stage. The experimental results show that our approach outperforms existing methods with the weighted average F1 scores about 90%. We also propose promising future directions to deploy this approach in real-life settings, such as applying multimodal learning using several supplementary sensors.

eess.SP↗

arXiv2020

An ensemble learning approach for software semantic clone detection

Code clone is a serious problem in software and has the potential to software defects, maintenance overhead, and licensing violations. Therefore, clone detection is important for reducing maintenance effort and improving code quality during software evolution. A variety of clone detection techniques have been proposed to identify similar code in software. However, few of them can efficiently detect semantic clones (functionally similar code without any syntactic resemblance). Recently, several deep learning based clone detectors are proposed to detect semantic clones. However, these approaches have high cost in data labelling and model training. In this paper, we propose a novel approach that leverages word embedding and ensemble learning techniques to detect semantic clones. Our evaluation on a commonly used clone benchmark, BigCloneBench, shows that our approach significantly improves the precision and recall of semantic clone detection, in comparison to a token-based clone detector, SourcererCC, and another deep learning based clone detector, CDLH.

cs.SE↗

arXiv2020

Directivity modulation of exciton emission using single dielectric nanospheres

Coupling emitters with nanoresonators is an effective strategy to control light emission at the subwavelength scale with high efficiency. Low-loss dielectric nanoantennas hold particular promise for this purpose, owing to their strong Mie resonances. Herein, we explore a highly miniaturized platform for the control of emission based on individual subwavelength Si nanospheres (SiNSs) to modulate the directional excitation and exciton emission of two-dimensional transition metal dichalcogenides (2D TMDs). A modified Mie theory for dipole-sphere hybrid systems is derived to instruct the optimal design for desirable modulation performance. Controllable forward-to-backward intensity ratios are experimentally validated in 532 nm laser excitation and 635 nm exciton emission from a monolayer WS2. Versatile light emission control along all device orientations is achieved for different emitters and excitation wavelengths, benefiting from the facile size control and isotropic shape of SiNSs. Simultaneous modulation of excitation and emission via a single SiNS at visible wavelengths significantly improves the efficiency and directivity of TMD exciton emission and leads to the potential of multifunctional integrated photonics. Overall, our work opens promising opportunities for nanophotonics and polaritonic systems, enabling efficient manipulation, enhancement and reconfigurability of light-matter interactions.

physics.optics↗

arXiv2020

Scalable NAS with Factorizable Architectural Parameters

Neural Architecture Search (NAS) is an emerging topic in machine learning and computer vision. The fundamental ideology of NAS is using an automatic mechanism to replace manual designs for exploring powerful network architectures. One of the key factors of NAS is to scale-up the search space, e.g., increasing the number of operators, so that more possibilities are covered, but existing search algorithms often get lost in a large number of operators. For avoiding huge computing and competition among similar operators in the same pool, this paper presents a scalable algorithm by factorizing a large set of candidate operators into smaller subspaces. As a practical example, this allows us to search for effective activation functions along with the regular operators including convolution, pooling, skip-connect, etc. With a small increase in search costs and no extra costs in re-training, we find interesting architectures that were not explored before, and achieve state-of-the-art performance on CIFAR10 and ImageNet, two standard image classification benchmarks.

cs.LG↗