arXiv subjects

Jia Yuan Yu

Publications and source records attributed to Jia Yuan Yu.

At least 19 recordsLinked to original sources

arXiv2026

Causal Inference on Stopped Random Walks in Online Advertising

We consider a causal inference problem frequently encountered in online advertising systems, where a publisher (e.g., Instagram, TikTok) interacts repeatedly with human users and advertisers by sporadically displaying to each user an advertisement selected through an auction. Each treatment corresponds to a parameter value of the advertising mechanism (e.g., auction reserve-price), and we want to estimate through experiments the corresponding long-term treatment effect (e.g., annual advertising revenue). In our setting, the treatment affects not only the instantaneous revenue from showing an ad, but also changes each user's interaction-trajectory, and each advertiser's bidding policy -- as the latter is constrained by a finite budget. In particular, each a treatment may even affect the size of the population, since users interact longer with a tolerable advertising mechanism. We drop the classical i.i.d. assumption and model the experiment measurements (e.g., advertising revenue) as a stopped random walk, and use a budget-splitting experimental design, the Anscombe Theorem, a Wald-like equation, and a Central Limit Theorem to construct confidence intervals for the long-term treatment effect.

stat.ML↗

arXiv2025

Introducing the O-Value: A Universal Standardization for Confusion-Matrix-Based Classification Performance Metrics

Many classification performance metrics exist, each suited to a specific application. However, these metrics often differ in scale and can exhibit varying sensitivity to class imbalance rates in the test set. As a result, it is difficult to use the nominal values of these metrics to evaluate, compare and monitor classification performances, especially when imbalance rates vary. To address this problem, we introduce the outperformance standardization (OPS) function, a universal standardization method for confusion-matrix-based classification performance (CMBCP) metrics. It maps any given metric to a common scale of $[0,1]$, while providing a clear and consistent interpretation. Specifically, the resulting OPS value (o-value) represents the percentile rank of the observed classification performance within a reference distribution of possible performances. This unified framework enables meaningful comparison and monitoring of classification performance across test sets with differing imbalance rates. We illustrate how o-values can be applied to a variety of commonly used classification performance metrics and demonstrate the utility and robustness of our method through experiments on real-world datasets spanning multiple classification applications.

stat.ML↗

arXiv2024

Error Analysis of Shapley Value-Based Model Explanations: An Informative Perspective

Shapley value attribution (SVA) is an increasingly popular explainable AI (XAI) method, which quantifies the contribution of each feature to the model's output. However, recent work has shown that most existing methods to implement SVAs have some drawbacks, resulting in biased or unreliable explanations that fail to correctly capture the true intrinsic relationships between features and model outputs. Moreover, the mechanism and consequences of these drawbacks have not been discussed systematically. In this paper, we propose a novel error theoretical analysis framework, in which the explanation errors of SVAs are decomposed into two components: observation bias and structural bias. We further clarify the underlying causes of these two biases and demonstrate that there is a trade-off between them. Based on this error analysis framework, we develop two novel concepts: over-informative and underinformative explanations. We demonstrate how these concepts can be effectively used to understand potential errors of existing SVA methods. In particular, for the widely deployed assumption-based SVAs, we find that they can easily be under-informative due to the distribution drift caused by distributional assumptions. We propose a measurement tool to quantify such a distribution drift. Finally, our experiments illustrate how different existing SVA methods can be over- or under-informative. Our work sheds light on how errors incur in the estimation of SVAs and encourages new less error-prone methods.

cs.AI↗

arXiv2022

On Unique Ergodicity Of Coupled AIMD Flows

The AIMD algorithm, which underpins the Transmission Control Protocol (TCP) for transporting data packets in communication networks, is perhaps the most successful control algorithm ever deployed. Recently, its use has been extended beyond communication networks, and successful applications of the AIMD algorithm have been reported in transportation, energy, and mathematical biology. A very recent development in the use of AIMD is its application in solving large-scale optimization and distributed control problems without the need for inter-agent communication. In this context, an interesting problem arises when multiple AIMD networks that are coupled in some sense (usually through a nonlinearity). The purpose of this note is to prove that such systems in certain settings inherit the ergodic properties of individual AIMD networks. This result has important consequences for the convergence of the aforementioned optimization algorithms. The arguments in the paper also correct conceptual and technical errors in [1].

math.DS↗

arXiv2022

Reward Modeling for Mitigating Toxicity in Transformer-based Language Models

Transformer-based language models are able to generate fluent text and be efficiently adapted across various natural language generation tasks. However, language models that are pretrained on large unlabeled web text corpora have been shown to suffer from degenerating toxic content and social bias behaviors, consequently hindering their safe deployment. Various detoxification methods were proposed to mitigate the language model's toxicity; however, these methods struggled to detoxify language models when conditioned on prompts that contain specific social identities related to gender, race, or religion. In this study, we propose Reinforce-Detoxify; A reinforcement learning-based method for mitigating toxicity in language models. We address the challenge of safety in language models and propose a new reward model that is able to detect toxic content and mitigate unintended bias towards social identities in toxicity prediction. The experiments demonstrate that the Reinforce-Detoxify method for language model detoxification outperforms existing detoxification approaches in automatic evaluation metrics, indicating the ability of our approach in language model detoxification and less prone to unintended bias toward social identities in generated content.

cs.CL↗

arXiv2021

Multi-resource allocation for federated settings: A non-homogeneous Markov chain model

In a federated setting, agents coordinate with a central agent or a server to solve an optimization problem in which agents do not share their information with each other. Wirth and his co-authors, in a recent paper, describe how the basic additive-increase multiplicative-decrease (AIMD) algorithm can be modified in a straightforward manner to solve a class of optimization problems for federated settings for a single shared resource with no inter-agent communication. The AIMD algorithm is one of the most successful distributed resource allocation algorithms currently deployed in practice. It is best known as the backbone of the Internet and is also widely explored in other application areas. We extend the single-resource algorithm to multiple heterogeneous shared resources that emerge in smart cities, sharing economy, and many other applications. Our main results show the convergence of the average allocations to the optimal values. We model the system as a non-homogeneous Markov chain with place-dependent probabilities. Furthermore, simulation results are presented to demonstrate the efficacy of the algorithms and to highlight the main features of our analysis.

math.OC↗

arXiv2021

Bias-Corrected Peaks-Over-Threshold Estimation of the CVaR

The conditional value-at-risk (CVaR) is a useful risk measure in fields such as machine learning, finance, insurance, energy, etc. When measuring very extreme risk, the commonly used CVaR estimation method of sample averaging does not work well due to limited data above the value-at-risk (VaR), the quantile corresponding to the CVaR level. To mitigate this problem, the CVaR can be estimated by extrapolating above a lower threshold than the VaR using a generalized Pareto distribution (GPD), which is often referred to as the peaks-over-threshold (POT) approach. This method often requires a very high threshold to fit well, leading to high variance in estimation, and can induce significant bias if the threshold is chosen too low. In this paper, we derive a new expression for the GPD approximation error of the CVaR, a bias term induced by the choice of threshold, as well as a bias correction method for the estimated GPD parameters. This leads to the derivation of a new estimator for the CVaR that we prove to be asymptotically unbiased. In a practical setting, we show through experiments that our estimator provides a significant performance improvement compared with competing CVaR estimators in finite samples. As a consequence of our bias correction method, it is also shown that a much lower threshold can be selected without introducing significant bias. This allows a larger portion of data to be be used in CVaR estimation compared with the typical POT approach, leading to more stable estimates. As secondary results, a new estimator for a second-order parameter of heavy-tailed distributions is derived, as well as a confidence interval for the CVaR which enables quantifying the level of variability in our estimator.

stat.ME↗

arXiv2020

Risk-Averse Action Selection Using Extreme Value Theory Estimates of the CVaR

In a wide variety of sequential decision making problems, it can be important to estimate the impact of rare events in order to minimize risk exposure. A popular risk measure is the conditional value-at-risk (CVaR), which is commonly estimated by averaging observations that occur beyond a quantile at a given confidence level. When this confidence level is very high, this estimation method can exhibit high variance due to the limited number of samples above the corresponding quantile. To mitigate this problem, extreme value theory can be used to derive an estimator for the CVaR that uses extrapolation beyond available samples. This estimator requires the selection of a threshold parameter to work well, which is a difficult challenge that has been widely studied in the extreme value theory literature. In this paper, we present an estimation procedure for the CVaR that combines extreme value theory and a recently introduced method of automated threshold selection by \cite{bader2018automated}. Under appropriate conditions, we estimate the tail risk using a generalized Pareto distribution. We compare empirically this estimation procedure with the commonly used method of sample averaging, and show an improvement in performance for some distributions. We finally show how the estimation procedure can be used in reinforcement learning by applying our method to the multi-arm bandit problem where the goal is to avoid catastrophic risk.

stat.ML↗

arXiv2020

Generating Embroidery Patterns Using Image-to-Image Translation

In many scenarios in computer vision, machine learning, and computer graphics, there is a requirement to learn the mapping from an image of one domain to an image of another domain, called Image-to-image translation. For example, style transfer, object transfiguration, visually altering the appearance of weather conditions in an image, changing the appearance of a day image into a night image or vice versa, photo enhancement, to name a few. In this paper, we propose two machine learning techniques to solve the embroidery image-to-image translation. Our goal is to generate a preview image which looks similar to an embroidered image, from a user-uploaded image. Our techniques are modifications of two existing techniques, neural style transfer, and cycle-consistent generative-adversarial network. Neural style transfer renders the semantic content of an image from one domain in the style of a different image in another domain, whereas a cycle-consistent generative adversarial network learns the mapping from an input image to output image without any paired training data, and also learn a loss function to train this mapping. Furthermore, the techniques we propose are independent of any embroidery attributes, such as elevation of the image, light-source, start, and endpoints of a stitch, type of stitch used, fabric type, etc. Given the user image, our techniques can generate a preview image which looks similar to an embroidered image. We train and test our propose techniques on an embroidery dataset which consist of simple 2D images. To do so, we prepare an unpaired embroidery dataset with more than 8000 user-uploaded images along with embroidered images. Empirical results show that these techniques successfully generate an approximate preview of an embroidered version of a user image, which can help users in decision making.

eess.IV↗

arXiv2020

The Convergence of Finite-Averaging of AIMD for Distributed Heterogeneous Resource Allocations

In several social choice problems, agents collectively make decisions over the allocation of multiple divisible and heterogeneous resources with capacity constraints to maximize utilitarian social welfare. The agents are constrained through computational or communication resources or privacy considerations. In this paper, we analyze the convergence of a recently proposed distributed solution that allocates such resources to agents with minimal communication. It is based on the randomized additive-increase and multiplicative-decrease (AIMD) algorithm. The agents are not required to exchange information with each other, but little with a central agent that keeps track of the aggregate resource allocated at a time. We formulate the time-averaged allocations over finite window size and model the system as a Markov chain with place-dependent probabilities. Furthermore, we show that the time-averaged allocations vector converges to a unique invariant measure, and also, the ergodic property holds.

math.OC↗

arXiv2019

A Price-Based Iterative Double Auction for Charger Sharing Markets

The unprecedented growth of demand for charging electric vehicles (EVs) calls for novel expansion solutions to today's charging networks. Riding on the wave of the proliferation of sharing economy, Airbnb-like charger sharing markets opens the opportunity to expand the existing charging networks without requiring costly and time-consuming infrastructure investments, yet the successful design of such markets relies on innovations at the interface between game theory, mechanism design, and large scale optimization. In this paper, we propose a price-based iterative double auction for charger sharing markets where charger owners rent out their under-utilized chargers to the charge-needing EV drivers. Charger owners and EV drivers form a two-sided market which is cleared by a price-based double auction. Chargers' locations, availability, and time unit costs as well as the EV drivers' time, distance constraints, and preferences are considered in the allocation and scheduling process. The goal is to compute social welfare maximizing allocations which benefits both charger owners and EV drivers and, in turn, ensure the continuous growth of the market. We prove that the proposed double auction is budget balanced, individually rational, and that it is a weakly dominant strategy for EV drivers and charger owners to truthfully report their charging time constraints. In addition, results from our computation study show that the double auction achieves on average 94% efficiency compared with the optimal solutions and scales well to larger problem instances.

cs.GT↗

arXiv2019

A Scheme for Dynamic Risk-Sensitive Sequential Decision Making

We present a scheme for sequential decision making with a risk-sensitive objective and constraints in a dynamic environment. A neural network is trained as an approximator of the mapping from parameter space to space of risk and policy with risk-sensitive constraints. For a given risk-sensitive problem, in which the objective and constraints are, or can be estimated by, functions of the mean and variance of return, we generate a synthetic dataset as training data. Parameters defining a targeted process might be dynamic, i.e., they might vary over time, so we sample them within specified intervals to deal with these dynamics. We show that: i). Most risk measures can be estimated using return variance; ii). By virtue of the state-augmentation transformation, practical problems modeled by Markov decision processes with stochastic rewards can be solved in a risk-sensitive scenario; and iii). The proposed scheme is validated by a numerical experiment.

cs.AI↗

arXiv2019

Variance-Based Risk Estimations in Markov Processes via Transformation with State Lumping

Variance plays a crucial role in risk-sensitive reinforcement learning, and most risk measures can be analyzed via variance. In this paper, we consider two law-invariant risks as examples: mean-variance risk and exponential utility risk. With the aid of the state-augmentation transformation (SAT), we show that, the two risks can be estimated in Markov decision processes (MDPs) with a stochastic transition-based reward and a randomized policy. To relieve the enlarged state space, a novel definition of isotopic states is proposed for state lumping, considering the special structure of the transformed transition probability. In the numerical experiment, we illustrate state lumping in the SAT, errors from a naive reward simplification, and the validity of the SAT for the two risk estimations.

cs.LG↗

arXiv2019

Adaptive and optimal online linear regression on $\ell^1$-balls

We consider the problem of online linear regression on individual sequences. The goal in this paper is for the forecaster to output sequential predictions which are, after $T$ time rounds, almost as good as the ones output by the best linear predictor in a given $\ell^1$-ball in $\\R^d$. We consider both the cases where the dimension~$d$ is small and large relative to the time horizon $T$. We first present regret bounds with optimal dependencies on $d$, $T$, and on the sizes $U$, $X$ and $Y$ of the $\ell^1$-ball, the input data and the observations. The minimax regret is shown to exhibit a regime transition around the point $d = \sqrt{T} U X / (2 Y)$. Furthermore, we present efficient algorithms that are adaptive, \ie, that do not require the knowledge of $U$, $X$, $Y$, and $T$, but still achieve nearly optimal regret bounds.

stat.ML↗

arXiv2018

Derandomized Distributed Multi-resource Allocation with Little Communication Overhead

We study a class of distributed optimization problems for multiple shared resource allocation in Internet-connected devices. We propose a derandomized version of an existing stochastic additive-increase and multiplicative-decrease (AIMD) algorithm. The proposed solution uses one bit feedback signal for each resource between the system and the Internet-connected devices and does not require inter-device communication. Additionally, the Internet-connected devices do not compromise their privacy and the solution does not dependent on the number of participating devices. In the system, each Internet-connected device has private cost functions which are strictly convex, twice continuously differentiable and increasing. We show empirically that the long-term average allocations of multiple shared resources converge to optimal allocations and the system achieves minimum social cost. Furthermore, we show that the proposed derandomized AIMD algorithm converges faster than the stochastic AIMD algorithm and both the approaches provide approximately same solutions.

eess.SY↗

arXiv2018

Distributed Algorithms for Internet-of-Things-enabled Prosumer Markets: A Control Theoretic Perspective

Internet-of-Things (IoT) enables the development of sharing economy applications. In many sharing economy scenarios, agents both produce as well as consume a resource; we call them prosumers. A community of prosumers agrees to sell excess resource to another community in a prosumer market. In this chapter, we propose a control theoretic approach to regulate the number of prosumers in a prosumer community, where each prosumer has a cost function that is coupled through its time-averaged production and consumption of the resource. Furthermore, each prosumer runs its distributed algorithm and takes only binary decisions in a probabilistic way, whether to produce one unit of the resource or not and to consume one unit of the resource or not. In the proposed approach, prosumers do not explicitly exchange information with each other due to privacy reasons, but little exchange of information is required for feedback signals, broadcast by a central agency. In the proposed approach, prosumers achieve the optimal values asymptotically. Furthermore, the proposed approach is suitable to implement in an IoT context with minimal demands on infrastructure. We describe two use cases; community-based car sharing and collaborative energy storage for prosumer markets. We also present simulation results to check the efficacy of the algorithms.

eess.SY↗

arXiv2018

Transition-based versus State-based Reward Functions for MDPs with Value-at-Risk

In reinforcement learning, the reward function on current state and action is widely used. When the objective is about the expectation of the (discounted) total reward only, it works perfectly. However, if the objective involves the total reward distribution, the result will be wrong. This paper studies Value-at-Risk (VaR) problems in short- and long-horizon Markov decision processes (MDPs) with two reward functions, which share the same expectations. Firstly we show that with VaR objective, when the real reward function is transition-based (with respect to action and both current and next states), the simplified (state-based, with respect to action and current state only) reward function will change the VaR. Secondly, for long-horizon MDPs, we estimate the VaR function with the aid of spectral theory and the central limit theorem. Thirdly, since the estimation method is for a Markov reward process with the reward function on current state only, we present a transformation algorithm for the Markov reward process with the reward function on current and next states, in order to estimate the VaR function with an intact total reward distribution.

cs.AI↗

arXiv2018

State-Augmentation Transformations for Risk-Sensitive Reinforcement Learning

In the framework of MDP, although the general reward function takes three arguments-current state, action, and successor state; it is often simplified to a function of two arguments-current state and action. The former is called a transition-based reward function, whereas the latter is called a state-based reward function. When the objective involves the expected cumulative reward only, this simplification works perfectly. However, when the objective is risk-sensitive, this simplification leads to an incorrect value. We present state-augmentation transformations (SATs), which preserve the reward sequences as well as the reward distributions and the optimal policy in risk-sensitive reinforcement learning. In risk-sensitive scenarios, firstly we prove that, for every MDP with a stochastic transition-based reward function, there exists an MDP with a deterministic state-based reward function, such that for any given (randomized) policy for the first MDP, there exists a corresponding policy for the second MDP, such that both Markov reward processes share the same reward sequence. Secondly we illustrate that two situations require the proposed SATs in an inventory control problem. One could be using Q-learning (or other learning methods) on MDPs with transition-based reward functions, and the other could be using methods, which are for the Markov processes with a deterministic state-based reward functions, on the Markov processes with general reward functions. We show the advantage of the SATs by considering Value-at-Risk as an example, which is a risk measure on the reward distribution instead of the measures (such as mean and variance) of the distribution. We illustrate the error in the reward distribution estimation from the direct use of Q-learning, and show how the SATs enable a variance formula to work on Markov processes with general reward functions.

cs.AI↗