This document discusses the results of using autoencoders to reduce the dimensionality of various datasets into two dimensions. It presents four case studies - a 20-dimensional mathematical example, a 40-dimensional mathematical example, a 35-dimensional truss structure, and a 6-story frame structure subject to lateral loads - where autoencoders were applied to each dataset and the results are displayed in three figures each to demonstrate the technique's feasibility for dimensionality reduction.
This document discusses the results of using autoencoders to reduce the dimensionality of various datasets into two dimensions. It presents four case studies - a 20-dimensional mathematical example, a 40-dimensional mathematical example, a 35-dimensional truss structure, and a 6-story frame structure subject to lateral loads - where autoencoders were applied to each dataset and the results are displayed in three figures each to demonstrate the technique's feasibility for dimensionality reduction.
This document discusses the results of using autoencoders to reduce the dimensionality of various datasets into two dimensions. It presents four case studies - a 20-dimensional mathematical example, a 40-dimensional mathematical example, a 35-dimensional truss structure, and a 6-story frame structure subject to lateral loads - where autoencoders were applied to each dataset and the results are displayed in three figures each to demonstrate the technique's feasibility for dimensionality reduction.
SECTION A: DIMENSIONALITY REDUCTION INTO TWO DIMENSIONS
Case Study I: A 20D Mathematical Example Three runs are done to ensure feasibility, shown in Figure 1-3. Case Study II: A 40D Mathematical Example Three runs are done to ensure feasibility, shown in Figure 4-6. Case Study III: A 35D Truss Structure Three runs are done to ensure feasibility, shown in Figure 7-9. Case Study IV: A 6-Story Frame Structure Subject to Lateral Loads with 25mm Drift Threshold Three runs are done to ensure feasibility, shown in Figure 10-12