Enhao Liu

Enhao Liu

Senior Healthcare Data Scientist at University of Maryland Medical System

Sitemap

A list of all the posts and pages found on the site. For you robots out there is an XML version available for digesting as well.

Pages

CV

Published:

Posts

Future Blog Post

less than 1 minute read

Published:

This post will show up by default. To disable scheduling of future posts, edit config.yml and set future: false.

Blog Post number 4

less than 1 minute read

Published:

This is a sample blog post. Lorem ipsum I can’t remember the rest of lorem ipsum and don’t have an internet connection right now. Testing testing testing this blog post. Blog posts are cool.

Blog Post number 3

less than 1 minute read

Published:

This is a sample blog post. Lorem ipsum I can’t remember the rest of lorem ipsum and don’t have an internet connection right now. Testing testing testing this blog post. Blog posts are cool.

Blog Post number 2

less than 1 minute read

Published:

This is a sample blog post. Lorem ipsum I can’t remember the rest of lorem ipsum and don’t have an internet connection right now. Testing testing testing this blog post. Blog posts are cool.

Blog Post number 1

less than 1 minute read

Published:

This is a sample blog post. Lorem ipsum I can’t remember the rest of lorem ipsum and don’t have an internet connection right now. Testing testing testing this blog post. Blog posts are cool.

portfolio

publications

Reward-based Monte Carlo-Bayesian Reinforcement Learning for Cyber Preventive Maintenance

Published in Computers & Industrial Engineering, 2018

Abstract: This article considers a preventive maintenance problem related to cyber security in universities. A Bayesian Reinforcement Learning (BRL) problem is formulated using limited data from scan results and intrusion detection system warnings. The median estimated learning time (MELT) measure is introduced to evaluate the speed at which a control system effectively eliminates parametric uncertainty and probability is concentrated on a single scenario. It is demonstrated that the Monte Carlo BRL with enhancements including Latin hypercube sampling (LHS) to generate scenarios, identical systems multi-task learning, and reward-based learning achieves shorter MELT values, i.e., “faster” learning, and improved objective values compared with alternatives in a numerical study. Rigorous results establish the optimality of the derived control strategies and the fact that optimal learning is possible under steady state assumptions. Also, the real-world case study of policies for patching Linux critical server cyber vulnerabilities generates insights including the potential to reduce expenditure per host by mandating compensating controls for critical vulnerabilities.

Forecasting Cyber Maintenance Costs with Improved Scan Analytics Using Simulation

Published in 2018 Winter Simulation Conference, 2018

Abstract: This article proposes a discrete event simulation model of an organization that maintains computer hosts and incurs several millions of dollars in maintenance and incident response costs. The common maintenance policy is referred to as “out-of-sight is out-of-mind” (OSOM) because the majority of hosts are absent from scans and ignored. Hosts are “dark” (absent) because they are not accessible (turned off or with restricted permissions). The proposed model is used to compare OSOM with alternatives including improved analytics that make dark host vulnerabilities visible. Findings clarify the apparent benefits of OSOM unless indirect costs for intrusions or improved policies are applied. Also, benefits from using Windows operating systems and improved policies are clarified including millions in expected savings (vs. Linux).

Discrete Event Simulation of Cyber Maintenance Policies According to Nested Birth And Death Processes

Published in 2019 Winter Simulation Conference, 2019

Abstract: This article proposes a novel discrete event simulation model for predicting cyber maintenance costs under multiple scenarios. In this study, our model of the evolution of computer hosts is similar to the Susceptible- Infected-Removed (S-I-R) epidemiological model. A double or “nested birth and death” construction is used for the hosts and the vulnerabilities on the hosts. The objectives of the model are to study the benefits and drawbacks of current scanning policy and maintenance policy, evaluate cost-effective alternatives, and investigate the significance of celebrity vulnerabilities.

Cyber Vulnerability Maintenance Policies That Address The Incomplete Nature of Inspection

Published in Applied Stochastic Models in Business and Industry, 2019

Abstract: In cybersecurity, incomplete inspection, resulting mainly from computers being turned off during the scan, leads to a challenge for scheduling maintenance actions. This article proposes the application of partially observable decision processes to derive cost‐effective cyber maintenance actions that minimize total costs. We consider several types of hosts having vulnerabilities at various levels of severity. The maintenance cost structure in our proposed model consists of the direct costs of maintenance actions in addition to potential incident costs associated with different security states. To assess the benefits of optimal policies obtained from partially observable Markov decision processes, we use real‐world data from a major university. Compared with alternative policies using simulations, the optimal control policies can significantly reduce expected maintenance expenditures per host and relatively quickly mitigate the most important vulnerabilities.

A Bayesian Spatiotemporal Nowcasting Model for Public Health Decision-Making and Surveillance

Published in American Journal of Epidemiology, 2022

Abstract: As coronavirus disease 2019 (COVID-19) spread through the United States in 2020, states began to set up alert systems to inform policy decisions and serve as risk communication tools for the general public. Many of these systems included indicators based on an assessment of trends in numbers of reported cases. However, when cases are indexed by date of disease onset, reporting delays complicate the interpretation of trends. Despite a foundation of statistical literature with which to address this problem, these methods have not been widely applied in practice. In this paper, we develop a Bayesian spatiotemporal nowcasting model for assessing trends in county-level COVID-19 cases in Ohio. We compare the performance of our model with the approach used in Ohio and the approach included in decision support materials from the Centers for Disease Control and Prevention. We demonstrate gains in performance while still retaining interpretability using our model. In addition, we are able to fully account for uncertainty in both the time series of cases and the reporting process. While we cannot eliminate all of the uncertainty in public health surveillance and subsequent decision-making, we must use approaches that embrace these challenges and deliver more accurate and honest assessments to policy-makers.

Optimal classification trees with leaf-branch and binary constraints

Published in Computers & Operations Research, 2024

Abstract: Using empirical models to predict whether sections within pipes have defects can save inspection costs and, potentially, avoid oil spills. Optimal Classification Tree (OCT) formulations offer potentially desirable combinations of interpretability and prediction accuracy on unseen pipes. Approaches based on powerful state-of-the-art OCT formulations have enabled researchers to solve decision tree problems optimally instead of using traditional sub-optimal greedy approaches. Yet, the recently proposed formulations also have limitations. Some of the most recent formulations require a large number of decision variables and constraints leading to computational inefficiencies. Previous formulations have optimal solutions with undesirable or invalid tree structures which may depend on the particular software implementation. Additionally, some formulations always grow a full tree even when desirable parsimonious tree options are available. This article proposes the Modified Optimal Classification Tree (M-OCT) formulation with novel leaf-branch-interaction constraints, which could stabilize the previous formulation and reduce the chance of invalid tree structures when generating optimal trees. By incorporating the idea of binary encoding of thresholds from a previous article, we reduce the total number of binary variables. We then extend M-OCT to construct a novel formulation called Binary Node Penalty Optimal Classification Tree (BNP-OCT) with binary splits and node complexity constraints, which support efficiency in standard branch-and-cut solvers and prevents the overfitting issue when learning the optimal tree models. We compare the proposed methods with alternatives including standard formulations using 15 standard data sets. In addition, we use 750 test cases to compare the computational stability of pre-existing formulations to those involving the proposed leaf-branch constraints. We demonstrate that the proposed formulation offers advantages in accuracy, computational efficiency, and structural stability. We also describe how the proposed methods are able to achieve 94% classification accuracy on balanced test sets for unseen pipes.

talks

teaching

Teaching experience 1

Undergraduate course, University 1, Department, 2014

This is a description of a teaching experience. You can use markdown like any other post.

Teaching experience 2

Workshop, University 1, Department, 2015

This is a description of a teaching experience. You can use markdown like any other post.