52 ISE Magazine | www.iise.org/ISEmagazine

Inside IISE Journals

research

How to efﬁciently manage

warehouse order picking activities

Warehouses are crucial for the efﬁciency of distribution

networks since they balance the variation in demand, con-

solidate various products and help with rapid response to

customer requests. In a classical warehouse with a picker-

to-parts system, pickers walk or ride along aisles and col-

lect the requested items. In this type of warehouse, batch

picking is a common practice where a set of orders, called

a batch, must be formed. A picker collects the items of the

orders belonging to a batch in a single tour with the objec-

tive of minimizing the total travel distance of the pickers.

On the other hand, the assignment of selected batches to

the pickers is nearly always made under time considerations.

Hence, the wave picking systems have important implica-

tions in delivery operations especially when the orders to

be collected constitute a truckload and the truck will not

be dispatched until all orders have been handed over by the

pickers.

This paper addresses the integration of order batching

and picker scheduling decisions by taking into account the

minimization of both the total travel time to collect all or-

ders by the pickers and the makespan of the orders. The in-

tegrated problem not only occurs naturally in wave picking

systems in which the latest picking time of orders becomes

the key performance metric, but also arises when there is a

limit on the picker operating time.

In their work, “Order Batching and Picker Scheduling in

Warehouse Order Picking” İbrahim Muter from Amazon

Web Services and Temel Öncan from the Galatasaray Uni-

versity in İstanbul propose a column generation-based exact

algorithm for the integrated problem. The novelty of the

proposed approach lies in the ability of efﬁciently solving

the integrated problem to optimality by designing a column

generation subproblem based on the set of batches.

The major ﬁnding is that the upper bound for the order

batching and scheduling problem obtained by solving the

order batching problem is very tight for instances with 50 or

more orders. For many instances, solving the order batch-

ing problem is sufﬁcient to ﬁnd the optimal solution for the

integrated problem.

CONTACT: Temel Öncan; ytoncan@gsu.edu.tr; Department of Industrial

Engineering, Galatasaray University, 34349 Ortaköy, İstanbul, Turkey

How to compress deep learning?

Try tensor decomposition

Deep learning represents a family of artiﬁcial intelligence

methods based on deep neural networks. It has attracted

much attention from industrial practitioners. It has been

This month we highlight two articles in IISE Transactions. The ﬁrst article looks into the question of how to efﬁciently

manage order picking activities in the warehouses. The authors focus on an integrated problem involving both order

batching and picker scheduling. Their research reveals that the upper bound for the combined order batching and

scheduling problem obtained by solving the order batching problem alone is very tight when there are a sufﬁcient

amount of orders. This insight suggests solving the order batching problem is sufﬁcient to ﬁnd the optimal solution for

the integrated decision problem. The second article asks, “How to compress deep learning?” The number of parameters

involved in a deep learning model increases too fast as the number of layers increases. Compressing deep learning

models and making them smaller will therefore make them run faster and more interpretable. The authors show that the

convolutional layers in deep learning can be reformulated and compressed by using tensor decomposition. These articles

will appear in the May 2022 issue (Volume 54, No. 5).

Temel ÖncanIbrahim Muter

April 2022 | ISE Magazine 53

applied to many ﬁelds, such

as computer vision, speech

recognition, anomaly detec-

tion, production prediction

and quality control, where

deep learning can achieve

high prediction or classiﬁ-

cation accuracy, sometimes

surpassing human expert

performance.

Although deep learning

is favorable in dealing with

large-scale and high-dimensional datasets due to its ac-

curacy, the huge number of parameters in deep learning

models will increase exponentially as the layers grow. This

limitation increases the model complexity, decays the fea-

ture interpretability and hinders its applications in devices

with limited memory space. Therefore, compressing deep

learning models is desirable for making them smaller, faster

and more interpretable.

In their paper, “Tensor Decomposition to Compress

Convolutional Layers in Deep Learning,” doctoral student

Yinan Wang, assistant professor Xiaowei Yue from Virginia

Tech and assistant professor Grace Guo from Rutgers Uni-

versity show that the convolutional layers in deep learning

can be reformulated and compressed by using tensor de-

composition to reduce model complexity while preserving

its accuracy.

The authors describe how the tensor decomposition is

incorporated into designing a novel convolutional layer in

deep learning. The newly proposed layer is highly modu-

larized and ready to be used with other layers. Both the

theoretical analysis and experimental results demonstrate

that the newly proposed layer uses fewer parameters, has

lower time complexity and maintains comparable perfor-

mance in image-based surface defect classiﬁcation. It also

has the ﬂexibility to adjust the number of parameters for

balancing the model complexity and its accuracy.

Furthermore, the proposed compressed layer can be gen-

erally used as a low-rank alternative to the convolutional

layers in broader applications. The idea of this work is also

extendable to compress other types of deep learning.

CONTACT: Xiaowei Yue; xwy@vt.edu; Grado Department of Industrial

and Systems Engineering, Virginia Tech, 1145 Perry Street, Blacksburg,

VA 24061

How artiﬁcial intelligence could

help pulmonologists and radiologists

diagnose lung cancer

Lung cancer is not only the most prevalent form of cancer,

but it also has a high mortality rate. To diagnose lung can-

cer at its early stages, computed tomography (CT) is the

most common way for pulmonologists and radiologists to

identify and analyze lung nodules by scanning the chest

area of patients. However, identifying malignant nodules,

which are rounded or irregular opacity at the lung region,

is very challenging because of nodule heterogeneity and the

lack of annotated lung nodule images.

Different pulmonologists and radiologists may have differ-

ent opinions and diagnostic results for a single lung nodule

because lung nodule analysis is highly dependent on subjective

interpretations and the radiologists’ preceding experiences. In

recent years, the development of computer-aided diagnosis

(CAD) has promoted lung nodule diagnosis by providing reli-

able diagnosis results and enhancing analysis effectiveness.

In their paper, “A Novel Self-Adaptive Convolutional

Neural Network Model Using Spatial Pyramid Pooling for

3D Lung Nodule Computer-Aided Diagnosis,”

This month we highlight two articles from IISE

Transactions on Healthcare Systems Engineering

(Volume 12, No. 1). In the first paper, researchers

studied lung nodule heterogeneity in diagnosing

malignancies from CT scans. In an effort to design

an effective nodule classification framework for

CAD systems, they proposed a deep learning

algorithm that identifies malignant and benign

lung nodules from 3D CT images in more effective

ways. In the second paper, the authors develop

an advanced epidemiological formulation in

combination with vaccine and treatment resources

allocation, namely epidemic-vaccination-logistics,

to support decision-making surrounding the

management of epidemic diseases. They present

a multistage stochastic programming framework

that evaluates various disease growth scenarios to

optimize distribution of treatment centers, medical

resources and vaccines.

Yinan Wang Xiaowei Yue

Grace Guo

54 ISE Magazine | www.iise.org/ISEmagazine

research

Qianqian Zhang and Sang Won Yoon of the State Univer-

sity of New York at Binghamton studied lung nodule hetero-

geneity and designed an effective nodule classiﬁcation frame-

work for CAD systems. They proposed a novel deep learning

algorithm that identiﬁes malignant and benign lung nodules

from 3D CT images in more effective ways.

3D lung CT images are collected by sorting 2D CT im-

ages based on their transverse positions on the craniocaudal

axis. It has been tested with 3D CT images from Lung Image

Database Consortium and Image Database Resource Initia-

tive (LIDC-IDRI), which is one of the largest lung datasets

publicly available for lung nodule screening images and nodule

annotations.

The experimental results show that the proposed method

improves lung nodule diagnosis accuracy up to 12.12%, with

fewer than 10% of the parameters that are involved in other

recent lung nodule diagnosis studies. This is a notable perfor-

mance achievement in terms of diagnostic sensitivity, showing

the algorithm’s capacity of recognizing malignant lung nod-

ules. In practice, the artiﬁcial intelligence and machine learn-

ing algorithms can offer complementary opinions in CAD

systems as a supportive tool for radiologists and physicians in

medical image interpretation, analysis and diagnosis.

CONTACT: Sang Won Yoon, Professor, Systems Science and Industrial

Engineering, Thomas J. Watson College of Engineering and Applied Science,

State University of New York at Binghamton, Binghamton, NY 13902

How healthcare decision-makers

can effectively respond to epidemic

under uncertainty and risk

Making resource-allocation decisions under a public health

emergency, such as the COVID-19 pandemic and Ebola virus

disease, is always challenging, especially given the uncertainty

of disease transmission among the human population. Health-

care professionals and government ofﬁcials often have to make

difﬁcult decisions before knowing the trajectory of the disease

progression and its impacts on their people. Since the evolu-

tion of an epidemic disease is quite dynamic and uncertain,

every decision should be made at the risk of experiencing di-

sastrous scenarios with many infections and deaths due to the

infectious disease.

In their paper, “Risk-Averse Multi-Stage Stochastic Pro-

gramming to Optimizing Vaccine Allocation and Treatment

Logistics for Effective Epidemic Response,” Xuecheng Yin,

a postdoctoral research associate at Yale University School

of Public Health and Esra Büyüktahtakın Toy, an associate

professor at New Jersey Institute of Technology, develop an

advanced epidemiological formulation in combination with

vaccine and treatment resources allocation, namely epidemic-

vaccination-logistics, to support decision-making surrounding

the management of epidemic diseases.

The authors present a multistage stochastic programming

framework that evaluates various disease growth scenarios

under the conditional value-at-risk (CVaR) to optimize the

distribution of treatment centers, medical resources and vac-

cines while minimizing the total expected number and risk of

infections, deaths and close contacts of infected people under

a limited budget.

The authors integrate ring vaccination strategy into an epi-

demics-logistics model in the paper. The model is formulated

under the uncertainty of vaccine supply and disease transmis-

sion rates and the potential risk of experiencing highly adverse

disease growth scenarios.

The authors incorporate human mobility into the model and

develop a new method to estimate the migration rate between

each spatial region while data on migration rates is not avail-

able. They demonstrate the mathematical optimization results

on the case of controlling the 2018-2020 Ebola virus disease in

the Democratic Republic of the Congo using real data.

The research shows that isolating and treating infected in-

dividuals are the most efﬁcient ways to slow the transmission

Qianqian Zhang

Esra Büyüktahtakın Toy Xuecheng Yin

April 2022 | ISE Magazine 55

of the Ebola Virus Disease, which transmits through physi-

cal contact. While vaccination is supplementary to primary

interventions, such as treatment and isolation, its delay could

cause an exponential increase in the number of infections and

deaths. Our ﬁndings also suggest using the limited budget for

vaccination in regions where the disease has just started, while

the model gives priority to building new Ebola treatment cen-

ters (ETCs) and treating infected people over vaccination in

regions with high initial infections. The results further show

that as the risk-averseness level increases, the budget allocated

to areas with the highest initial infection level is decreased by

moving the ETCs and treatment budget to neighboring lo-

cations under the risk of getting infections. In addition, the

results indicate that vaccine acceptance rates affect the optimal

vaccine-allocation policy only at the initial stages of the vac-

cine rollout when the vaccine supply is tight.

CONTACT: Esra Büyüktahtakın Toy, Mechanical and Industrial Engineering,

NJIT; esratoy@njit.edu; web.njit.edu/~esratoy; University Heights, Mechani-

cal and Industrial Engineering Center, Suite 204, Newark, NJ 07102; Xuech-

eng Yin, xuecheng.yin@yale.edu; School of Public Health, Yale University,

350 George St., New Haven, CT 06511

Yu Ding is the Mike and Sugar Barnes Professor of Industrial

and Systems Engineering at Texas A&M University and Associ-

ate Director for Research Engagement at the Texas A&M Institute

of Data Science. He is editor-in-chief of IISE Transactions and

a fellow of IISE.

Oguzhan Alagoz is a professor in the Department of Industrial and

Systems Engineering at the University of Wisconsin-Madison. He

is editor-in-chief of IISE Transactions on Healthcare Systems

Engineering and a fellow of IISE.

IIISE Transactions (link.iise.org/iisetransactions) is IISE’s ﬂagship

research journal and is published monthly. It aims to foster

exchange among researchers and practitioners in the industrial

engineering community by publishing papers that are grounded

in science and mathematics and motivated by engineering

applications.

IISE Transactions on Healthcare Systems Engineering (link.iise.org/

iisetransactions_healthcare) is a quarterly, refereed journal that

publishes papers about the application of industrial engineering tools

and techniques to healthcare systems.

To subscribe, call (800) 494-0460 or (770) 449-0460.

About the journals

Advanced Excel Tools for ISEs ● Honing Your FMEA Skills ● The Basics of Simulation

Regression Refresher ● Systems Engineering Tools for Process Improvement ● and more …

iise.org/microlearning

The quickest, most cost-efficient CEUs around!