Graduate Students Seminar
Location
Mathematics/Psychology : 106
Date & Time
October 2, 2024, 11:00 am – 11:50 am
Description
Session Chair: | Vishal Subedi |
Discussant: | Dr. Mathew |
Speaker 1: Ji Li
- Title
- Parallelism of Four-Parameter Logistic Bioassay Models: Equivalence Testing for Parameter Differences
- Abstract
- Parallelism is a prerequisite assumption that the test product behaves like a dilution or concentration of the reference product. It is the basis for defining the relative potency of a test product to the reference standard. Once the parallelism between test and reference curves is established, the relative potency will be constant at any effective response levels. For the 4-parameter logistic bioassay models, multivariate equivalence tests based on the ratio of the parameters was proposed to establish parallelism. In order to determine the tolerance limits, one needs to rely on a complicated procedure with modern technology. As an alternative, we propose the use of parameter difference-based equivalence tests using historical data to simplify the process of margin determination. This method addresses the issue of reduced power as precision decreases associated with the use of fixed margins. The effectiveness of this approach is demonstrated through a comprehensive simulation study.
Speaker 2: Nathan Tamiru
- Title
- Exploring AlignedReID and ST-RNNs: Neural Network Approaches for Person Re-Identification and Next-Location Prediction
- Abstract
- In this study, we focus on a detailed examination of two contemporary methodologies: AlignedReID and Spatio-Temporal Recurrent Neural Networks (ST-RNNs). These methods leverage the combined capabilities of Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) to address challenges related to person re-identification and next-location prediction. Our investigation involves a thorough analysis of these implementations, encompassing their design intricacies and performance evaluations through different metrics and testing procedures. The significance of these advanced approaches lies in their ability to effectively track individuals across varied camera angles and provide predictive insights into their future locations. The practical uses of this approach cover a wide range of areas, including improving surveillance systems, strengthening security protocols, and optimizing decision-making processes for greater insight.