Monday, November 9 9-11am
Recent advancements in BARON
Presented by: Dr. Yash Puranik and Aphrodite Kishi
In this workshop, we will discuss the recent developments in the BARON project and exciting new features in the latest BARON release. We will use extensive computational results to demonstrate BARON’s performance over various problems and problem types. We will provide demos using BARON through its different interfaces, including Pyomo (Python), GAMS, and JuMP. We will also give a summary of recent BARON applications.
A deductive approach to text analytics through construction and validation of categorization models.
In this session, we will introduce participants to a powerful deductive approach in text analytics using categorization models. Such an approach allows analysts and researchers to identify the degree to which specific concepts or topics are present within a large volume of documents. This approach requires one to build dictionaries with multiple categories and subcategories containing words, word patterns, phrases and rules that are then used to extract specific content of interest. Sentiment analysis is a well-known example of such an approach, but we will see that there are many more applications of categorization models. Dr. Normand Peladeau will demonstrate how to build, validate, and deploy a categorization dictionary to answer specific research questions or test hypotheses.
Applying AnyLogic Simulation to Solve Various Business Challenges
Presented by: Arash Mahdavi, AI Program Lead, AnyLogic North America and Jeffrey Wirtz, Program Support Specialist, AnyLogic North America
In this workshop, we will discuss how you can leverage the unique features of AnyLogic simulation software and AnyLogic Cloud to solve your business challenges.
We will demonstrate applications of simulation in various domains such as Manufacturing, Healthcare, Supply Chain, Railway, Oil and Gas, Human Resources, Pedestrian movement, and Road Traffic. During the workshop, we will discuss different simulation methodologies (Discrete Event, Agent Based, System Dynamics) and how you can use them in any combination (Multi-method modeling).
We will also showcase AnyLogic Cloud – the most advanced cloud solution existing today for simulation – and go through its typical use cases. Our topics will include: integration of simulation into custom analytical workflows, scalable high-performance computing, instant delivery of models to clients and users, the creation of custom web interfaces for simulation models, and much more. We will also go through the AnyLogic Cloud’s open API and introduce AnyLogic Private Cloud for organizations with strict security guidelines.
Tuesday, November 10 9-11am
A Brief Tour of GAMS: From Basic Modeling to Model Deployment with Graphical UIs and Use of Dedicated Compute Hardware
Presented by: Atharv Bhosekar, PhD, Senior Engineer; Stefan Mann, PhD, Technical Sales Engineer; Frederik Proske, Lead Developer MIRO and ENGINE
The General Algebraic Modeling System (GAMS) allows modelers to create optimization-based decision support applications. In this workshop, we will first focus on model development with GAMS: What’s in a model, how to solve different problem types (linear, mixed-integer, non-linear) with GAMS, how to switch solvers, and how to separate the model code from input data using GDX. We will then present GAMS MIRO, our system for creating interactive user interfaces for GAMS models. We will demonstrate how to create a user interface, how to run the model in MIRO, compare multiple scenarios, create graphics of the output data, and the use of our powerful pivot table feature. Lastly, we will discuss our new deployment option GAMS ENGINE, which allows running GAMS models on dedicated servers, freeing the end-user PCs from the computational workload. We will show how to use ENGINE from GAMS Studio and from MIRO, how user management works in ENGINE, and how to schedule multiple jobs and collect the results.
Calling all OR Heroes: Learn How to Model and Deploy Solutions in 75% Less Time with FICO® Xpress Insight
Presentedy by: Baykal Hafizoglu
Join our hands-on workshop to experience how FICO® Xpress Insight helps you read data in any format from any source, integrates with your own machine learning and solvers (or Xpress Solver), enables collaboration with business users, deploys decision support or automated solutions in the cloud or on premises —and does it all in 75% less time.
- Get the overview of the great new features of FICO® Xpress Solver, Mosel, Workbench, and Insight
- Learn how you can rapidly deploy Python models with FICO Xpress Insight
- Discover the new View Designer, which reduces GUI development times from minutes to seconds
- Experience real-world business examples in which developers integrate their own solver(s) with the powerful Xpress Insight platform
- See how the flexible, free-to-use Xpress Mosel modeling and programming language allows you to make your solver available to thousands of other Mosel users
All attendees will receive a link to the complimentary FICO® Xpress Optimization Community License.
New Innovations: Cloud Computing, Real-Time Scheduling, Industry 4.0, and More
Presented by: Ryan Welch Luttrell – Solutions Engineer (Simio LLC) and Adam Sneath – Solutions Engineer (Simio LLC)
With Simio leveraging the cloud computing power of Microsoft Azure to support your most demanding applications, its compatibility with Schneider Electric’s Wonderware to allow detailed production scheduling with real-time data and risk analysis, and ability to schedule and reschedule in real-time, Simio is leading the way in Industry 4.0 and creating a digital factory. Outside our immense technology partner advances, we have great new features, application areas and capabilities! Come explore an overview of the new Simio experience and see why we are always “Forward Thinking.”
Wednesday, November 11 9-11am
An Interactive Workshop: Managing AI and Optimization Project Risks with “The Princeton 20”
Presented by: Irv Lustig, PhD, Optimization Principal, Princeton Consultants and Patricia Randall, PhD, Director, Princeton Consultants
Analytics practitioners and team leaders are invited to an interactive, in-depth discussion of the core challenges to bringing AI and optimization from the insight stage into deployed production. Irv Lustig and Patricia Randall will present “The Princeton 20,” a framework to score and manage common risks of for AI and optimization projects, sharing examples from past projects at Princeton Consultants and elsewhere.
Each attendee is encouraged to have at least one current or past project in mind, and comment on and share risk mitigation tips and techniques as Irv and Patricia lead a group discussion through 10 business factors and 10 technical factors. For each, participants will be asked to describe their selected projects with respect to that factor.
For 40 years, Princeton Consultants has had the privilege of working with innovators across a wide range of industries to improve their decision-making using advanced analytics. We have found consistent and repeatable factors that govern whether a given analytics project will make it into successful production. These factors also apply to the latest AI projects. In this workshop, you will learn:
- 10 business risk factors and 10 technical risk factors to monitor in AI and optimization projects.
- How to score project risk factors.
- Tips and techniques to overcome obstacles and deploy solutions.
Recent Developments in the Gurobi Optimizer 9.1
Presented by: Zonghao Gu, PhD, CTO and CO-Founder of Gurobi Optimization; Tobias Achterberg, PhD, Vice President of Research & Development, Gurobi Optimization; Ed Klotz, PhD, Senior Mathematical Optimization Specialist, Gurobi Optimization
During this session, we will provide an overview of the new Gurobi 9.1 release. We will highlight some of the new features included in this release such as the enhancements to the Python API, and also talk about the speed-ups that we achieved in our MIP, LP and QP algorithms. In particular, we will cover our Irreducible Infeasible Set of constraints (IIS) algorithm and its improvements, the new MIP heuristic for models with extremely difficult LP relaxations, and we will discuss some of the techniques that resulted in the huge speed-up of our non-convex MIQCP solver.
Advances in the New Releases of LINGO, LINDO API, and What’sBest!
Presented by: Linus Schrage, CEO and founder of LINDO Systems
Learn how the extensive enhancements in LINDO API, LINGO, and What’sBest! make it easier to quickly formulate and solve your optimization problems. Find out about the latest performance improvements and new and enhanced features and interfaces. The new releases offer an order of magnitude performance improvement in solving “long skinny” LP’s with many more variables than constraints. Speed in solving LPs with multiple cores using the concurrent solver has been significantly improved.
The new Parameter Tuning Tool helps boost performance by finding the best parameter settings for a given set of models. Performance on non-convex quadratic models has been significantly improved using the Global Solver’s substantially expanded ability to recognize and exploit convexity. Among the new features, support has been added for Multiple/Lexico objectives. Interfaces to support most popular modeling languages have been added and enhanced, as well as general purpose frontends such as R, Excel, and more. Powerful but easy to use new features have been added to LINGO to integrate your optimization models more smoothly into Excel workbooks.
LINGO has expanded the file types that it can read and write and enhanced its ability to work with dynamic sets. The Stochastic optimization capabilities utilize improved management of large numbers of scenarios. What’sBest! offers more robust support to the wide range of mathematical functions in Excel.
Thursday, November 12 9-11am
FlexSim in Academe
Presented by: Allen Greenwood, Ph.D., P.E.(retired) – Professor Emeritus of Industrial and Systems Engineering at Mississippi State University and Simulation Education Specialist at FlexSim Software Products, Inc.
Join us in discussing how FlexSim simulation software has been used to enhance teaching, research, and outreach. After a brief introduction to FlexSim, Dr. Greenwood shares his nearly 15 years of experience with using the software as an integral part of his courses and applied research, and as a link with industry. He provides examples of designing and managing operations systems in a variety of domains, such as manufacturing, healthcare, transportation, logistics, supply chains, etc. The examples are drawn from his experiences as a faculty member in universities in both the U.S. and abroad. The workshop provides time for discussion of these topics among the participants and more in-depth exploration of FlexSim “problem solved.” as needed based on participant requests.
Optimization with MATLAB®
Join us for a hands-on workshop using MATLAB® Online™ to try out the new problem-based approach that makes it easier to define and solve optimization problems with MATLAB. We’ll work through a variety of examples to learn how to
- Construct objectives and constraints from expressions of optimization variables using MATLAB operators and functions
- Define sets of variables and constraints using arrays indexed by numbers and strings
- Include black-box functions
- Apply an automatically selected solver
- Use automatic differentiation for nonlinear problems
- Build an app around the problem, using the drag-and-drop interface of App Designer
We’ll also briefly present how you can include machine learning and how you can deploy your applications to both enterprise and embedded systems.
To participate in the hands-on exercises, before the workshop:
Analyzing Multidimensional Data with Ease
Presented by: Kevin Potcner
As data becomes more available across almost every industry, analyses of that data require more than the 1-sample and 2-sample inferential analysis techniques traditionally taught in the standard one and two semester statistics courses. For students to be prepared for today’s job market, their analysis toolbox needs to be expanded to include tools that are much more powerful capable of finding patterns and relationships across a large number of variables. The presenter will illustrate how JMP’s interactive point-and-click environment makes it easy to explore highly dimensional data intuitively with dynamic visualizations and advanced data mining and predictive modeling capabilities.
Friday, November 13 9-11am
1) IBM Large Data Forecasting Toolkit (40min)
Forecasting for large numbers of related time series is a common need across many domains. E.g., in supply chains it may be necessary to forecast demand for up to billions of product-store time series. In process industries production processes can be instrumented with thousands of sensors, producing large numbers of time series with complex relationships. Handling the scale for fine-grained forecasting while supporting a variety of forecasting components (including deep learning techniques) and their automated selection and evaluation is a challenging task, not yet fully addressed in existing systems. We present a toolkit for this purpose and describe its design, components, features, and application along with benchmarks across data sets for different techniques (including DL and traditional forecasting techniques).
2) Smarter Resource and Operations Management (40min)
AI powered analytics and automation is critical for realizing enhanced productivity and investment return. In this talk, we present SROM, a toolkit suite that provides out-of-the-box purpose built machine learning and optimization pipelines targeted at core functions of the industry. This helps accelerate and operationalize AI towards building custom applications with data from assets and processes. The Intelligent Assets and Operations toolkit is focused on advanced techniques for anomaly detection, failure pattern analysis, cohort analysis, and enhanced pipelines for rapid authoring, training and deployment of Industry 4.0 applications. The second toolkit in the suite, referred to as Situational Awareness toolkit, brings together multi-model data and domain semantics to maintain an active semantic representation of the cyber-physical plant, reflecting the state and health of operations for actionable awareness. The Intelligent System and Optimization toolkit provides optimization techniques which couple machine learning (including regression, deep learning) with novel optimization techniques for providing data driven decision making. We provide a description of each of the toolkits along with use cases that demonstrate the application of these capabilities in real-world situations.
3) IBM Decision Optimization – New Highlights (40min)
The first part of the workshop will focus on the IBM Decision Optimization offerings. The upcoming version of CPLEX Optimization Studio will offer not only performance improvements, but also a number of new features, including:
– Black-box optimization in CP Optimizer.
– Database support in OPL
– A new CPLEX MIP emphasis mode with extreme focus on primal performance
We will describe these features and their usage. We will also highlight the recent evolutions of Cloud Pak for Data, the IBM platform for the development and deployment of Machine Learning and Decision Optimization models.
“Solving Business Problems with SAS Analytics and OPTMODEL”
Presented by: Rob Pratt, Senior R&D Manager, Ed Hughes, Principal Product Manager, and members of the SAS Operations Research Center of Excellence
SAS offers diverse analytic capabilities, including data integration, statistical analysis, data and text mining, machine learning, artificial intelligence, forecasting, optimization, scheduling, and simulation. The OPTMODEL procedure from SAS provides you with a full-featured optimization modeling language, access to a diverse set of solvers, and the ability to create and use customized solution algorithms. SAS analytic capabilities are also available through the cloud-enabled, open design of SAS® Viya®. You can program in SAS or in other languages—Python, Lua, Java, and R.
We’ll explore analytical and optimization case studies drawn directly from our work with SAS users in a wide range of industries. These case studies demonstrate PROC OPTMODEL’s power and versatility in building and solving optimization models and illustrate how deeply optimization integrates with the full array of analytics provided by SAS. We especially emphasize projects in which SAS has assisted in and
helped guide responses to the ongoing coronavirus / COVID-19 pandemic.