Thursday, February 2
Georgiana Haldeman, Colgate University
“Effective and Scalable CS Education”
With enrollments at a record high, computer science (CS) departments have a need for effective and scalable teaching techniques that is greater than ever. Some concepts and skills in CS are very challenging, such as runtime analysis and value vs. reference semantics. In this talk, we present our research on two pedagogical devices for these challenging concepts, and two additional solutions developed to make an existing scalable practice more effective, both for learning and teaching. These solutions extend autograders, enabling them to automatically provide formative feedback to students as well as a summary of the most common errors. These solutions are complementary: the first provides hints written by the instructor, while the second provides feedback on the structuring of students’ code.
Friday, February 3
Saad Hassan, Rochester Institute of Technology
“The Use of Automatic Sign Recognition to Support Look-up Technologies for Sign Language Learners: A Human-Computer Interaction Perspective”
Despite some prior research and commercial systems, if someone who is viewing sign language sees an unfamiliar sign, it remains a difficult task to look up its meaning. Advances in sign-language recognition technology have enabled the design of search systems for sign look-up in dictionaries, by allowing users to generate a query by submitting a video of themselves performing the word or by submitting a video segment containing the unknown sign. In this talk, we present our research investigating the user experience of sign-language search systems. We describe the design of two iteratively implemented sign look-up systems to support task-based experimental studies with sign language users and investigate settings of key performance and design variables. We also discuss future research in this area, on designing technologies to support bi-directional ASL learning including both receptive and expressive skills.
Thursday, February 16
"Wrangling Trustworthy COVID-19 Medical Research Data"
One of the challenges of working with large-scale structured datasets is how we access information. Medical paper datasets, such as ones focused on COVID-19, have the added difficulty that when querying a specific topic, a number of entries can be missed because authors use a variety of different table schemas. In order to create a profiler which successfully summarizes each topic of interest, the way in which we wrangle training data is very important. In this talk we will go over the process used by our research lab to make trustworthy COVID-19 information more accessible.
Friday, February 17
Matthew Rodriguez, Lehigh University
"Solving Graph Problems Efficiently on Modern Hardware with a Relaxation-First Priority Queue"
Graphs are a flexible and powerful abstraction which are capable of representing many different problem spaces. Many different graph algorithms powered by priority queues have been developed to solve various problems. However, in the era of multi-core CPUs, strict priority queues have an unavoidable contention bottleneck on the head which seriously limits parallelism and therefore performance. This can be addressed through relaxation, a strategy by which a priority queue does not always return the absolute highest priority element. While this can cause work to be wasted through reprocessing of nodes, the gains that result from increased parallelism often more than make up for that. Furthermore, in many real-world workloads, the degree of key repetition is so high that an extreme degree of relaxation can be tolerated with almost no cost.
Thus, we propose a design for a relaxed concurrent priority queue consisting of a linked list of vectors which utilize atomic integer operations for concurrent access. An index is maintained for fast insertion, and contention on the head is reduced by extracting an entire vector at once, and popping elements out of it one-by-one until it is empty. Teams of threads, organized by hardware considerations, share extracted chunks. We examine preliminary experimental results against the recent relaxed priority queue SprayList, and next steps for the project.
Tuesday, February 21
“Joins in Databases”
Joining tables in a database is important because it enables us to combine data from multiple sources into a single, comprehensive result set.
Without joins, we would be limited to retrieving data only from a single table, which would not always provide the complete picture.
Joins are used for a variety of purposes, including:
- Data aggregation: By combining data from multiple tables, we can aggregate and summarize information in a way that would not be possible from a single table.
- Data analysis: Joining tables can allow us to analyze data from different perspectives and answer more complex questions that would not be possible with data from a single table.
- Data normalization: Joining tables can help enforce data normalization, a database design technique that reduces redundancy and improves data integrity.
Data integration: Joining tables from different databases or data sources can help integrate data and provide a unified view of information.
By the end of this lecture, you would have learned how to use different types of JOINs effectively.
Tuesday, April 25
Alissa Crans, Loyola Marymount University
"The Mathematics of Tiling"
Interest in tiling patterns dates back to 4000 BCE. Together, we’ll construct two recent ones: the Conway Pinwheel and — hot off the press! — the March 2023 discovery of an aperiodic monotile called the “hat.” Join us as we discuss the math behind these tilings and create sidewalk art outside of Tome!
Tome 117 or Tome Outdoor Classroom
Tuesday, April 25
Mathematics & Computer Science Majors Dinner
Professor Barry Tesman will present "Can you hear me now? Exploring the mathematics of telephone numbers!"
We'll induct new members into Pi Mu Epsilon and Upsilon Pi Epsilon and present departmental prizes and awards
HUB Social Hall
Dinner provided by Dining Services
Wednesday, May 10
End of Year Picnic
Rain Location: Rector Atrium
Panera lunch provided