arXiv subjects

Yuan Yi

Publications and source records attributed to Yuan Yi.

6 recordsLinked to original sources

arXiv2023

MLOps Spanning Whole Machine Learning Life Cycle: A Survey

Google AlphaGos win has significantly motivated and sped up machine learning (ML) research and development, which led to tremendous ML technical advances and wider adoptions in various domains (e.g., Finance, Health, Defense, and Education). These advances have resulted in numerous new concepts and technologies, which are too many for people to catch up to and even make them confused, especially for newcomers to the ML area. This paper is aimed to present a clear picture of the state-of-the-art of the existing ML technologies with a comprehensive survey. We lay out this survey by viewing ML as a MLOps (ML Operations) process, where the key concepts and activities are collected and elaborated with representative works and surveys. We hope that this paper can serve as a quick reference manual (a survey of surveys) for newcomers (e.g., researchers, practitioners) of ML to get an overview of the MLOps process, as well as a good understanding of the key technologies used in each step of the ML process, and know where to find more details.

cs.LG↗

arXiv2022

Consecutive Piatetski-Shapiro primes based on the Hardy-Littlewood conjecture

The Piatetski-Shapiro sequences are of the form ${\mathcal{N}}^{(c)} := (\lfloor n^c \rfloor)_{n=1}^\infty$ with $c > 1, c \not\in \mathbb{N}$. In this paper, we study the distribution of pairs $(p, p^{\#})$ of consecutive primes such that $p \in {\mathcal{N}}^{(c_1)}$ and $p^{\#} \in {\mathcal{N}}^{(c_2)}$ for $c_1, c_2 > 1$ and give a conjecture with the prime counting functions of the pairs $(p, p^{\#})$. We give a heuristic argument to support this prediction which relies on a strong form of the Hardy-Littlewood conjecture. Moreover, we prove a proposition related to the average of singular series with a weight of a complex exponential function.

math.NT↗

arXiv2020

Data-Driven Stochastic Optimization for Power Grids Scheduling under High Wind Penetration

To address the environmental concern and improve the economic efficiency, the wind power is rapidly integrated into smart grids. However, the inherent uncertainty of wind energy raises operational challenges. To ensure the cost-efficient, reliable and robust operation, it is critically important to find the optimal decision that can correctly and rigorously hedge against all sources of uncertainty. In this paper, we propose data-driven stochastic unit commitment (SUC) to guide the power grids scheduling. Specifically, given the finite historical data, the posterior predictive distribution is developed to quantify the wind power prediction uncertainty accounting for both inherent stochastic uncertainty of wind power generation and input model estimation error. For complex power grid systems, a finite number of scenarios is used to estimate the expected cost in the planning horizon. To further control the impact of finite sampling error induced by using the sample average approximation (SAA), we propose a parallel computing based optimization solution methodology, which can quickly find the reliable optimal unit commitment decision hedging against various sources of uncertainty. The empirical study over six-bus and 118-bus systems demonstrates that our approach can provide more efficient and robust performance than the existing deterministic and stochastic unit commitment approaches.

eess.SY↗

arXiv2019

Global-Local Metamodel Assisted Two-Stage Optimization via Simulation

To integrate strategic, tactical and operational decisions, the two-stage optimization has been widely used to guide dynamic decision making. In this paper, we study the two-stage stochastic programming for complex systems with unknown response estimated by simulation. We introduce the global-local metamodel assisted two-stage optimization via simulation that can efficiently employ the simulation resource to iteratively solve for the optimal first- and second-stage decisions. Specifically, at each visited first-stage decision, we develop a local metamodel to simultaneously solve a set of scenario-based second-stage optimization problems, which also allows us to estimate the optimality gap. Then, we construct a global metamodel accounting for the errors induced by: (1) using a finite number of scenarios to approximate the expected future cost occurring in the planning horizon, (2) second-stage optimality gap, and (3) finite visited first-stage decisions. Assisted by the global-local metamodel, we propose a new simulation optimization approach that can efficiently and iteratively search for the optimal first- and second-stage decisions. Our framework can guarantee the convergence of optimal solution for the discrete two-stage optimization with unknown objective, and the empirical study indicates that it achieves substantial efficiency and accuracy.

math.OC↗

arXiv2011

A note on the moments of Kloosterman sums

In this note, we deduce an asymptotic formula for even power moments of Kloosterman sums based on the important work of N. M. Katz on Kloosterman sheaves. In a similar manner, we can also obtain an upper bound for odd power moments. Moreover, we shall give an asymptotic formula for odd power moments of absolute Kloosterman sums. Consequently, we find that there are infinitely many $a\bmod p$ such that $S(a,1;p)\gtrless0$ as $p\rightarrow+\infty.$

math.NT↗

arXiv2009

On character sums over flat numbers

Let $q\geqslant2$ be an integer, $χ$ be any non-principal character mod $q$, and $H=H(q)\leqslant q.$ In this paper the authors prove some estimates for character sums of the form \[\mathcal{W}(χ,H;q)=\sum_{n\in\mathscr{F}(H)}χ(n),\] where \[\mathscr{F}(H)=\left\{n\in\mathbb{Z}|(n,q)=1,1\leqslant n,\bar{n}\leqslant q, |n-\bar{n}|\leqslant H\},\] $\bar{n}$ is defined by $n\bar{n}\equiv1\pmod q.$

math.NT↗