arXiv subjects

Yang Li

Publications and source records attributed to Yang Li.

At least 973 records · Page 54Linked to original sources

arXiv2023

A Unified and Efficient Coordinating Framework for Autonomous DBMS Tuning

Recently using machine learning (ML) based techniques to optimize modern database management systems has attracted intensive interest from both industry and academia. With an objective to tune a specific component of a DBMS (e.g., index selection, knobs tuning), the ML-based tuning agents have shown to be able to find better configurations than experienced database administrators. However, one critical yet challenging question remains unexplored -- how to make those ML-based tuning agents work collaboratively. Existing methods do not consider the dependencies among the multiple agents, and the model used by each agent only studies the effect of changing the configurations in a single component. To tune different components for DBMS, a coordinating mechanism is needed to make the multiple agents cognizant of each other. Also, we need to decide how to allocate the limited tuning budget among the agents to maximize the performance. Such a decision is difficult to make since the distribution of the reward for each agent is unknown and non-stationary. In this paper, we study the above question and present a unified coordinating framework to efficiently utilize existing ML-based agents. First, we propose a message propagation protocol that specifies the collaboration behaviors for agents and encapsulates the global tuning messages in each agent's model. Second, we combine Thompson Sampling, a well-studied reinforcement learning algorithm with a memory buffer so that our framework can allocate budget judiciously in a non-stationary environment. Our framework defines the interfaces adapted to a broad class of ML-based tuning agents, yet simple enough for integration with existing implementations and future extensions. We show that it can effectively utilize different ML-based agents and find better configurations with 1.4~14.1X speedups on the workload execution time compared with baselines.

cs.DB↗

arXiv2023

Moiré Magnetic Exchange Interactions in Twisted Magnets

Besides moiré superlattice, twisting can also generate moiré magnetic exchange interactions (MMEIs) in van der Waals magnets. However, due to the extreme complexity and twist-angle-dependent sensitivity, all existing models fail to capture the MMEIs, preventing the understanding of MMEIs-induced new physics. Here, we develop a microscopic moiré spin Hamiltonian that enables the effective description of MMEIs via a sliding-mapping approach in twisted magnets, as demonstrated in twisted bilayer CrI3. Unexpectedly, we discover that the emergence of MMEIs can create an unprecedented magnetic skyrmion bubble (SkB) with non-conversed helicity, named as moiré-type SkB, representing a unique spin texture solely generated by MMEIs and ready to be detected under the current experimental conditions. Importantly, the size and population of SkBs can be finely controlled by twist angle, a key step for skyrmion-based quantum computing and information storage. Furthermore, we reveal that the MMEIs can be effectively manipulated by the substrate-induced interfacial Dzyaloshinskii-Moriya interaction, modulating the twist-angle-dependent magnetic phase diagram, which solves the outstanding disagreements between prior theories and experiments and verifies our theory.

cond-mat.mtrl-sci↗

arXiv2023

Factorized Blank Thresholding for Improved Runtime Efficiency of Neural Transducers

We show how factoring the RNN-T's output distribution can significantly reduce the computation cost and power consumption for on-device ASR inference with no loss in accuracy. With the rise in popularity of neural-transducer type models like the RNN-T for on-device ASR, optimizing RNN-T's runtime efficiency is of great interest. While previous work has primarily focused on the optimization of RNN-T's acoustic encoder and predictor, this paper focuses the attention on the joiner. We show that despite being only a small part of RNN-T, the joiner has a large impact on the overall model's runtime efficiency. We propose to utilize HAT-style joiner factorization for the purpose of skipping the more expensive non-blank computation when the blank probability exceeds a certain threshold. Since the blank probability can be computed very efficiently and the RNN-T output is dominated by blanks, our proposed method leads to a 26-30% decoding speed-up and 43-53% reduction in on-device power consumption, all the while incurring no accuracy degradation and being relatively simple to implement.

eess.AS↗

arXiv2023

Attention-based Graph Convolution Fusing Latent Structures and Multiple Features for Graph Neural Networks

We present an attention-based spatial graph convolution (AGC) for graph neural networks (GNNs). Existing AGCs focus on only using node-wise features and utilizing one type of attention function when calculating attention weights. Instead, we propose two methods to improve the representational power of AGCs by utilizing 1) structural information in a high-dimensional space and 2) multiple attention functions when calculating their weights. The first method computes a local structure representation of a graph in a high-dimensional space. The second method utilizes multiple attention functions simultaneously in one AGC. Both approaches can be combined. We also propose a GNN for the classification of point clouds and that for the prediction of point labels in a point cloud based on the proposed AGC. According to experiments, the proposed GNNs perform better than existing methods. Our codes open at https://github.com/liyang-tuat/SFAGC.

cs.CV↗

arXiv2023

Self-Calibrated Atom-Interferometer Gyroscope by Modulating Atomic Velocities

Atom-interferometer gyroscopes have attracted much attention for their potential superior long-term stability and extremely low drift. For such high precision instrument, a self-calibration to achieve an absolute rotation measurement is highly demanded. Here we propose and demonstrate a self-calibration of the atomic gyroscope. The calibration is realized by using the detuning of laser frequency to control the atomic velocity thus to modulate the scale factor of the gyroscope. The modulation determines the order and the initial phase of the interference stripe, thus eliminates the ambiguity caused by the periodicity of the interferometric signal. The calibration method is verified by measuring the Earth's rotation. Long-term stable and self-calibrated atom-interferometer gyros can find important applications in the fields of fundamental physics and long-time navigation.

physics.atom-ph↗

arXiv2023

Compact multi-channel radio frequency pulse sequence generator with fast switching capability for cold atom interferometers

Cold atom interferometers have matured to a powerful tool in fundamental physics research, and they are currently on their way from realizations in the laboratory to applications in the real world. The radio frequency (RF) generator is an indispensable component to control lasers and then manipulate atoms. We developed a highly compact RF generator for fast switching and sweeping frequencies/amplitudes of atomic interference pulse sequences. Multi-channel RF signals are generated by using a field-programmable gate array (FPGA) to control eight direct digital synthesizers (DDSs). We further proposed and demonstrated a method of preloading the parameters of all RF pulse sequences to the DDS registers before the execution of the pulse sequences, which eliminates the data transfer between the FPGA and DDSs to change RF signals and thus sharply shortens the delay of frequency switching when the pulse sequences are running. The characterized performance shows the generated RF signals achieve 119 ns frequency switching delay, and 40 dB harmonic rejection ratio. The generated RF pulse sequences are applied to a cold atom-interferometer gyroscope, and the contrast of atomic interference fringes reaches 38%. This compact multi-channel generator with the fast frequency/amplitude switching or/and sweeping capability has beneficial applications for the real-world atom interferometers.

physics.atom-ph↗

arXiv2023

Detecting the Most Probable High Dimensional Transition Pathway Based on Optimal Control Theory

Many natural systems exhibit phase transition where external environmental conditions spark a shift to a new and sometimes quite different state. Therefore, detecting the behavior of a stochastic dynamic system such as the most probable transition pathway, has made sense.We consider stochastic dynamic systems driven by Brownian motion. Based on variational principle and Onsager-Machlup action functional theory, the variational problem is transformed into deterministic optimal control problem. Different from traditional variational method, optimal control theory can handle high dimensional problems well. This paper describes the following three systems, double well system driven by additive noise, Maire-Stein system driven by multiplicative noise, and Nutrient-Phytoplankton-Nooplanktonr system.For numerical computation, based on Pontryagin's Maximum Principle, we adopt method of successive approximations to compute the optimal solution pathway, which is the most probable transition pathway in the sense of Onsager - Machlup.

math.OC↗

arXiv2023

Re-recognition of the ideal gas and real gas

Ideal gas is the most fundamental and simple system in thermodynamics, which has extensive applications in energy research and engineering. By reviewing the physical concept of ideal gas, it is found that the current understanding of ideal gas is still inappropriate and ambiguous, making it challenging to reveal the essential difference between ideal and real gases. Therefore, the macro and microscopic definitions and properties of ideal gas need to be reconceptualized. On the microscopic level, an ideal gas is a hypothetical collection of classical masses in irregular motion, which does not involve quantum mechanics; on the macroscopic level, any real gas cannot strictly obey the equation of state of ideal gas. Moreover, the heat capacity is a constant which should be an endogenous attribute of ideal gas model in order to unify the macroscopic and microscopic definitions of ideal gas. Finally, according to the law of heat capacity, the real gas can be divided into three categories: far-ideal, near-ideal, and quasi-ideal. This makes it easier to perform thermodynamic calculations with the assistance of ideal gas. Among them, the far-ideal gas heat capacity is a function of temperature and pressure, the near-ideal gas heat capacity is a single-valued function of temperature, and the quasi-ideal gas heat capacity is a constant.

physics.class-ph↗

arXiv2023

Practical Continuous-variable Quantum Key Distribution with Feasible Optimization Parameters

Continuous-variable quantum key distribution (CV-QKD) offers an approach to achieve a potential high secret key rate (SKR) in metropolitan areas. There are several challenges in developing a practical CV-QKD system from the laboratory to the real world. One of the most significant points is that it is really hard to adapt different practical optical fiber conditions for CV-QKD systems with unified hardware. Thus, how to improve the performance of practical CV-QKD systems in the field without modification of the hardware is very important. Here, a systematic optimization method, combining the modulation variance and error correction matrix optimization, is proposed to improve the performance of a practical CV-QKD system with a restricted capacity of postprocessing. The effect of restricted postprocessing capacity on the SKR is modeled as a nonlinear programming problem with modulation variance as an optimization parameter, and the selection of an optimal error correction matrix is studied under the same scheme. The results show that the SKR of a CV-QKD system can be improved by 24% and 200% compared with previous frequently used optimization methods theoretically with a transmission distance of 50 km. Furthermore, the experimental results verify the feasibility and robustness of the proposed method, where the achieved optimal SKR achieved practically deviates <1.6% from the theoretical optimal value. Our results pave the way to deploy high-performance CV-QKD in the real world.

quant-ph↗

arXiv2023

Unusual Nonlinear Optical Responses in Layered Ferroelectric Niobium Oxide Dihalides: Origin and Manipulation

Realization of large and highly tunable second-order nonlinear optical (NLO) responses, e.g., second-harmonic generation (SHG) and bulk photovoltaic effect (BPVE), is critical for developing modern optical and optoelectronic devices. Very recently, the two-dimensional van der Waals ferroelectric NbOX2 (X = Cl, Br or I) are discovered to exhibit unusually large and anisotropic SHG. However, the physical origin and possible tunability of NLO responses in NbOX2 remain to be unclear. In this article, we reveal that the large SHG in NbOCl2 is dominated by the synergy between large transition dipole moment and band-nesting-induced large intensity of electron-hole pairs. Remarkably, the NbOCl2 can exhibit dramatically different strain-dependent BPVE under different polarized light, originating from the interesting light-polarization-dependent orbital transition. Importantly, we successfully achieve a reversible ferroelectric-to-antiferroelectric phase transition via controlling ambient temperature or external pressure, accompanied by the greatly tunable NLO responses. Furthermore, we discover that the evolutions of SHG and BPVE in NbOX2 with variable X obey different rules. Our study provides a deep understanding on the novel NLO physics in NbOX2 and establishes great external-field tunability for device applications.

cond-mat.mtrl-sci↗

arXiv2023

Spotlight: Mobile UI Understanding using Vision-Language Models with a Focus

Mobile UI understanding is important for enabling various interaction tasks such as UI automation and accessibility. Previous mobile UI modeling often depends on the view hierarchy information of a screen, which directly provides the structural data of the UI, with the hope to bypass challenging tasks of visual modeling from screen pixels. However, view hierarchies are not always available, and are often corrupted with missing object descriptions or misaligned structure information. As a result, despite the use of view hierarchies could offer short-term gains, it may ultimately hinder the applicability and performance of the model. In this paper, we propose Spotlight, a vision-only approach for mobile UI understanding. Specifically, we enhance a vision-language model that only takes the screenshot of the UI and a region of interest on the screen -- the focus -- as the input. This general architecture of Spotlight is easily scalable and capable of performing a range of UI modeling tasks. Our experiments show that our model establishes SoTA results on several representative UI tasks and outperforms previous methods that use both screenshots and view hierarchies as inputs. Furthermore, we explore multi-task learning and few-shot prompting capacities of the proposed models, demonstrating promising results in the multi-task learning direction.

cs.CV↗

arXiv2023

Applications of integrated optical microcombs

Optical microcombs represent a new paradigm for generating laser frequency combs based on compact chip-scale devices, which have underpinned many modern technological advances for both fundamental science and industrial applications. Along with the surge in activity related to optical micro-combs in the past decade, their applications have also experienced rapid progress, not only in traditional fields such as frequency synthesis, signal processing, and optical communications, but also in new interdisciplinary fields spanning the frontiers of light detection and ranging (LiDAR), astronomical detection, neuromorphic computing, and quantum optics. This paper reviews the applications of optical microcombs. First, an overview of the devices and methods for generating optical microcombs is provided, which are categorized into material platforms, device architectures, soliton classes, and driving mechanisms. Second, the broad applications of optical microcombs are systematically reviewed, which are categorized into microwave photonics, optical communications, precision measurements, neuromorphic computing, and quantum optics. Finally, the current challenges and future perspectives are discussed.

physics.optics↗

arXiv2023

Enabling Conversational Interaction with Mobile UI using Large Language Models

Conversational agents show the promise to allow users to interact with mobile devices using language. However, to perform diverse UI tasks with natural language, developers typically need to create separate datasets and models for each specific task, which is expensive and effort-consuming. Recently, pre-trained large language models (LLMs) have been shown capable of generalizing to various downstream tasks when prompted with a handful of examples from the target task. This paper investigates the feasibility of enabling versatile conversational interactions with mobile UIs using a single LLM. We designed prompting techniques to adapt an LLM to mobile UIs. We experimented with four important modeling tasks that address various scenarios in conversational interaction. Our method achieved competitive performance on these challenging tasks without requiring dedicated datasets and training, offering a lightweight and generalizable approach to enable language-based mobile interaction.

cs.HC↗

arXiv2023

Transfer Learning for Bayesian Optimization: A Survey

A wide spectrum of design and decision problems, including parameter tuning, A/B testing and drug design, intrinsically are instances of black-box optimization. Bayesian optimization (BO) is a powerful tool that models and optimizes such expensive "black-box" functions. However, at the beginning of optimization, vanilla Bayesian optimization methods often suffer from slow convergence issue due to inaccurate modeling based on few trials. To address this issue, researchers in the BO community propose to incorporate the spirit of transfer learning to accelerate optimization process, which could borrow strength from the past tasks (source tasks) to accelerate the current optimization problem (target task). This survey paper first summarizes transfer learning methods for Bayesian optimization from four perspectives: initial points design, search space design, surrogate model, and acquisition function. Then it highlights its methodological aspects and technical details for each approach. Finally, it showcases a wide range of applications and proposes promising future directions.

cs.LG↗

arXiv2023

Anti-chiral edge states based on photonic Floquet lattices

Photonic Floquet lattices provide an excellent platform for manipulating different topologically protect-ed edge states. However, anti-chiral edge states have not been discussed much in Floquet lattices. Here, we propose a waveguide structure by combining two honeycomb Floquet photonic lattices with oppo-site rotation directions. In this structure, we find that the anti-chiral edge states have the same trans-mission direction on two parallel body edges. With an increasing modulation phase difference between the two sublattices in one direction, the width of the band gap becomes smaller and the robustness of the edge states becomes weaker. Interestingly, the transmission speed is also controlled by the phase difference. In addition to their relevance for the topological properties of the Floquet lattice system, these results may be applied to multi-channel optical switches, optical functional devices, and in other fields.

physics.optics↗

arXiv2023

Several classes of Galois self-orthogonal MDS codes and related applications

Let $q=p^h$ be a prime power and $e$ be an integer with $0\leq e\leq h-1$. $e$-Galois self-orthogonal codes are generalizations of Euclidean self-orthogonal codes ($e=0$) and Hermitian self-orthogonal codes ($e=\frac{h}{2}$ and $h$ is even). In this paper, we propose two general methods to construct $e$-Galois self-orthogonal (extended) generalized Reed-Solomon (GRS) codes. As a consequence, eight new classes of $e$-Galois self-orthogonal (extended) GRS codes with odd $q$ and $2e\mid h$ are obtained. Based on the Galois dual of a code, we also study its punctured and shortened codes. As applications, new $e'$-Galois self-orthogonal maximum distance separable (MDS) codes for all possible $e'$ satisfying $0\leq e'\leq h-1$, new $e$-Galois self-orthogonal MDS codes via the shortened codes, and new MDS codes with prescribed dimensional $e$-Galois hull via the punctured codes are derived. Moreover, some new $\sqrt{q}$-ary quantum MDS codes with lengths greater than $\sqrt{q}+1$ and minimum distances greater than $\frac{\sqrt{q}}{2}+1$ are obtained.

cs.IT↗

arXiv2023

DivBO: Diversity-aware CASH for Ensemble Learning

The Combined Algorithm Selection and Hyperparameters optimization (CASH) problem is one of the fundamental problems in Automated Machine Learning (AutoML). Motivated by the success of ensemble learning, recent AutoML systems build post-hoc ensembles to output the final predictions instead of using the best single learner. However, while most CASH methods focus on searching for a single learner with the best performance, they neglect the diversity among base learners (i.e., they may suggest similar configurations to previously evaluated ones), which is also a crucial consideration when building an ensemble. To tackle this issue and further enhance the ensemble performance, we propose DivBO, a diversity-aware framework to inject explicit search of diversity into the CASH problems. In the framework, we propose to use a diversity surrogate to predict the pair-wise diversity of two unseen configurations. Furthermore, we introduce a temporary pool and a weighted acquisition function to guide the search of both performance and diversity based on Bayesian optimization. Empirical results on 15 public datasets show that DivBO achieves the best average ranks (1.82 and 1.73) on both validation and test errors among 10 compared methods, including post-hoc designs in recent AutoML systems and state-of-the-art baselines for ensemble learning on CASH problems.

cs.LG↗

arXiv2023

Metric SYZ conjecture for certain toric Fano hypersurfaces

We prove the metric version of the SYZ conjecture for a class of Calabi-Yau hypersurfaces inside toric Fano manifolds, by solving a variational problem whose minimizer may be interpreted as a global solution of the real Monge-Ampère equation on certain polytopes. This does not rely on discrete symmetry.

math.DG↗