Taking the Person Seriously: Ethically Aware IS Research in the Era of Reinforcement Learning-Based Personalization

2023 | Journal of the Association for Information Systems | Citations: 0

Authors: Greene, Travis; Shmueli, Galit; Ray, Soumya

Abstract: Advances in reinforcement learning and implicit data collection on large-scale ... Expand

Abstract: Advances in reinforcement learning and implicit data collection on large-scale commercial platforms mark the beginning of a new era of personalization aimed at the adaptive control of human user environments. We present five emergent features of this new paradigm of personalization that endanger persons and societies at scale and analyze their potential to reduce personal autonomy, destabilize social and political systems, and facilitate mass surveillance and social control, among other concerns. We argue that current data protection laws, most notably the European Union's General Data Protection Regulation, are limited in their ability to adequately address many of these issues. Nevertheless, we believe that IS researchers are well-situated to engage with and investigate this new era of personalization. We propose three distinct directions for ethically aware reinforcement learning-based personalization research uniquely suited to the strengths of IS researchers across the sociotechnical spectrum. Collapse

Topics: usability personalization legal context General Data Protection Regulation information privacy law

Methods: reinforcement learning computational algorithm simulation language model experiment

Theories: social sciences theory

Preventive Care Now or Pay Later? A Personalized Medicine Approach for Healthcare Management

2023 | International Conference on Information Systems | Citations: 0

Authors: Guo, Tianjian; Bardhan, Indranil; Wen, Wen

Abstract: ... Expand

Abstract: Collapse

Topics: health information system mobile system information exchange Python price management

Methods: simulation reinforcement learning Markov decision process PyTorch deep learning

Scientific Workflows in IoT Environments: A Data Placement Strategy Based on Heterogeneous Edge-Cloud Computing

2022 | ACM Transactions on Management Information Systems | Citations: 0

Authors: Du, Xin; Tang, Songtao; Lu, Zhihui; Gai, Keke; Wu, Jie; Hung, Patrick C. K.

Abstract: In Industry 4.0 and Internet of Things (IoT) environments, the heterogeneous edg ... Expand

Abstract: In Industry 4.0 and Internet of Things (IoT) environments, the heterogeneous edge-cloud computing paradigm can provide a more proper solution to deploy scientific workflows compared to cloud computing or other traditional distributed computing. Owing to the different sizes of scientific datasets and the privacy issue concerning some of these datasets, it is essential to find a data placement strategy that can minimize data transmission time. Some state-of-the-art data placement strategies combine edge computing and cloud computing to distribute scientific datasets. However, the dynamic distribution of newly generated datasets to appropriate datacenters and exiting the spent datasets are still a challenge during workflows execution. To address this challenge, this study not only constructs a data placement model that includes shared datasets within the individual and among multiple workflows across various geographical regions, but also proposes a data placement strategy (DYM-RL-DPS) based on algorithms of two stages. First, during the build-time stage of workflows, we use the discrete particle swarm optimization algorithm with differential evolution to pre-allocate initial datasets to proper datacenters. Then, we reformulate the dynamic datasets distribution problem as a Markov decision process and provide a reinforcement learning–based approach to learn the data placement strategy in the runtime stage of scientific workflows. Through using the heterogeneous edge-cloud computing architecture to simulate IoT environments, we designed comprehensive experiments to demonstrate the superiority of DYM-RL-DPS. The results of our strategy can effectively reduce the data transmission time as compared to other strategies. Collapse

Topics: cloud computing mobile system internet of things privacy innovation management

Methods: computational algorithm experiment reinforcement learning simulation Markov decision process

Deep Reinforcement Learning for Supply Chain Synchronization

2022 | HICSS | Citations: 0

Authors: Jackson, Ilya

Abstract: Supply chain synchronization can prevent the “bullwhip effect” and significantly ... Expand

Abstract: Supply chain synchronization can prevent the “bullwhip effect” and significantly mitigate ripple effects caused by operational failures. This paper demonstrates how deep reinforcement learning agents based on the proximal policy optimization algorithm can synchronize inbound and outbound flows if end-toend visibility is provided. The paper concludes that the proposed solution has the potential to perform adaptive control in complex supply chains. Furthermore, the proposed approach is general, task unspecific, and adaptive in the sense that prior knowledge about the system is not required. Collapse

Topics: supply chain management open source

Methods: experiment computational algorithm reinforcement learning artificial neural network operations research

Theories: control theory

Efficient Representation for Electric Vehicle Charging Station Operations using Reinforcement Learning

2022 | HICSS | Citations: 0

Authors: Kwon, Kyung-Bin; Zhu, Hao

Abstract: Effectively operating an electric vehicle charging station (EVCS) is crucial for ... Expand

Abstract: Effectively operating an electric vehicle charging station (EVCS) is crucial for enabling the rapid transition of electrified transportation. By utilizing the flexibility of EV charging needs, the EVCS can reduce the total electricity cost for meeting the EV demand. To solve this problem using reinforcement learning (RL), the dimension of state/action spaces unfortunately grows with the number of EVs, which becomes very large and time-varying. This dimensionality issue affects the efficiency and convergence performance of generic RL algorithms. To this end, we advocate to develop aggregation schemes for state/action according to the emergency of EV charging, or its laxity. A least-laxity first (LLF) rule is used to consider only the total charging power of the EVCS, while ensuring the feasibility of individual EV schedules. In addition, we propose an equivalent state aggregation that can guarantee to attain the same optimal policy. Using the proposed aggregation scheme, the policy gradient method is applied to find the best parameters of a linear Gaussian policy. Numerical tests have demonstrated the performance improvement of the proposed representation approaches in increasing the total reward and policy efficiency over existing approximation-based method. Collapse

Methods: computational algorithm reinforcement learning literature study case study Markov decision process

Intra-day Dynamic Rescheduling under Patient No-shows

2022 | HICSS | Citations: 0

Authors: Shetty, Aditya; Groenevelt, Harry; Tilson, Vera

Abstract: Existing work on appointment scheduling assumes that appointment times cannot be ... Expand

Abstract: Existing work on appointment scheduling assumes that appointment times cannot be updated once they have been assigned. However, advances in communication technology and the adoption of online (as opposed to in-person) appointments make it possible for appointments to be flexible. In this paper, we describe an intra-day dynamic rescheduling model that adjusts upcoming appointments based on observed no-shows. We formulate the problem as a Markov Decision Process in order to compute the optimal pre-day schedule and the optimal policy to update the schedule for every scenario of no-shows. We also propose an alternative formulation based on the idea of ’atomic’ actions that can solve for the optimal policy more efficiently. Based on a numerical study, we estimate that using intra-day dynamic rescheduling can lead to a 5-7% decrease in expected cost when compared to static scheduling. Collapse

Topics: Python

Methods: Markov decision process experiment literature study optimization model

Should We Outlaw Ransomware Payments?

2021 | HICSS | Citations: 0

Authors: Dey, Debabrata; Lahiri, Atanu

Abstract: In recent times, there has been an upsurge in ransomware attacks, where an attac ... Expand

Abstract: In recent times, there has been an upsurge in ransomware attacks, where an attacker encrypts a user’s files and then demands a ransom in exchange for the decryption key. While paying the ransom allows the user to quickly unlock the locked files and avoid potentially larger losses, it also strengthens the hands of the attacker and increases the chance of a future attack. We study this dilemma of the victims and the externality posed by their actions using a game-theoretic model on top of a Markov decision process. The resulting equilibrium leads to several interesting insights such as that legally prohibiting ransom payments may not always have the desired economic effects—in some cases, a ban may be effective in addressing the economic externality but, in others, it could reduce public welfare. Our findings have important implications for policymakers who are currently debating legislation that, if enacted, will outlaw ransom payments to attackers. Collapse

Topics: unintended consequence ransomware legal context

Methods: Markov decision process

Theories: game theory

Managing Data Quality of the Data Warehouse: A Chance-Constrained Programming Approach

2021 | Information Systems Frontiers | Citations: 0

Authors: Liu, Qi; Feng, Gengzhong; Tayi, Giri Kumar; Tian, Jun

Abstract: To make informed decisions, managers establish data warehouses that integrate mu ... Expand

Abstract: To make informed decisions, managers establish data warehouses that integrate multiple data sources. However, the outcomes of the data warehouse-based decisions are not always satisfactory due to low data quality. Although many studies focused on data quality management, little effort has been made to explore effective data quality control strategies for the data warehouse. In this study, we propose a chance-constrained programming model that determines the optimal strategy for allocating the control resources to mitigate the data quality problems of the data warehouse. We develop a modified Artificial Bee Colony algorithm to solve the model. Our work contributes to the literature on evaluation of data quality problem propagation in data integration process and data quality control on the data sources that make up the data warehouse. We use a data warehouse in the healthcare organization to illustrate the model and the effectiveness of the algorithm. Collapse

Topics: data quality data warehouse data integration resource allocation quality management

Methods: computational algorithm optimization model sensitivity analysis experiment Monte Carlo simulation

Decisions for information or information for decisions? Optimizing information gathering in decision-intensive processes

2021 | Decision Support Systems | Citations: 0

Authors: Voorberg, S.; Eshuis, R.; van Jaarsveld, W.; van Houtum, G.J.

Abstract: Decision-intensive business processes are performed by decision makers who gathe ... Expand

Abstract: Decision-intensive business processes are performed by decision makers who gather different pieces of infor­ mation to reach the process objective: a final decision of high quality, for instance, the final price of a quote or the diagnosis of a failure of a hightech machine, as a result of an information-gathering process with minimum costs and efforts. Gathering all possible pieces of information results in high quality decisions, but also yields high costs and efforts. Therefore, decision makers require decision support to determine which information to gather to make the best final decision. This paper introduces an approach that supports a decision maker in the continuous trade-off between the effective acquisition of more information and cost-efficient decision making. The approach uses a well-defined modeling notation for decision-intensive processes, CMMN, and links it to a standard optimization technique, Markov Decision Processes. The approach calculates an optimal informationgathering solution, such that the expected result of the main decision minus the process cost for collecting in­ formation is optimized. The approach uses the solution to configure a run-time recommendation tool for the decision maker. The approach is flexible and allows that a decision maker ignores the advice; it then continues to offer recommendations in the subsequent states. We show the feasibility and effectiveness of our approach on a real-world quote process. Collapse

Topics: decision making decision support Python declarative language business process management

Methods: decision tree classification computational algorithm design science Markov decision process business process modeling

Combining Spatial Optimization and Multi-Agent Temporal Difference Learning for Task Assignment in Uncertain Crowdsourcing

2020 | Information Systems Frontiers | Citations: 0

Authors: Sun, Yong; Tan, Wenan

Abstract: In recent years, spatial crowdsourcing has emerged as an important new framework ... Expand

Abstract: In recent years, spatial crowdsourcing has emerged as an important new framework, in which each spatial task requires a set of right crowd-workers in the near vicinity to the target locations. Previous studies have focused on spatial task assignment in the static crowdsourcing environment. These algorithms may achieve local optimality by neglecting the uncertain features inherent in real-world crowdsourcing environments, where workers may join or leave during run time. Moreover, spatial task assignment is more complicated when large-scale crowd-workers exist in crowdsourcing environments. The large-scale nature of task assignments poses a significant challenge to uncertain spatial crowdsourcing. In this paper, we propose a novel algorithm combining spatial optimization and multi-agent temporal difference learning (SMATDL). The combination of grid-based optimization and multi-agent learning can achieve higher adaptability and maintain greater efficiency than traditional learning algorithms in the face of large-scale crowdsourcing problems. The SMATDL algorithm decomposes the uncertain crowdsourcing problem into numerous sub-problems by means of a grid-based optimization approach. In order to adapt to the change in the large-scale environment, each agent utilizes temporal difference learning to handle its own spatial region optimization in online crowdsourcing. As a result, multiple agents in SMATDL collaboratively learn to optimize their efforts in accomplishing the global assignment problems efficiently. Through extensive experiments, we illustrate the effectiveness and efficiency of our proposed algorithms on the experimental data sets. Collapse

Topics: crowdsourcing Java Python social network decision making

Methods: computational algorithm experiment optimization model case study reinforcement learning