Showing 1 - 2 of 2 Items
Teaching Computers to Teach Themselves: Synthesizing Training Data based on Human-Perceived Elements
Date: 2019-05-01
Creator: James Little
Access: Open access
- Isolation-Based Scene Generation (IBSG) is a process for creating synthetic datasets made to train machine learning detectors and classifiers. In this project, we formalize the IBSG process and describe the scenarios—object detection and object classification given audio or image input—in which it can be useful. We then look at the Stanford Street View House Number (SVHN) dataset and build several different IBSG training datasets based on existing SVHN data. We try to improve the compositing algorithm used to build the IBSG dataset so that models trained with synthetic data perform as well as models trained with the original SVHN training dataset. We find that the SVHN datasets that perform best are composited from isolations extracted from existing training data, leading us to suggest that IBSG be used in situations where a researcher wants to train a model with only a small amount of real, unlabeled training data.
A Comprehensive Survey on Functional Approximation
Date: 2022-01-01
Creator: Yucheng Hua
Access: Open access
- The theory of functional approximation has numerous applications in sciences and industry. This thesis focuses on the possible approaches to approximate a continuous function on a compact subset of R2 using a variety of constructions. The results are presented from the following four general topics: polynomials, Fourier series, wavelets, and neural networks. Approximation with polynomials on subsets of R leads to the discussion of the Stone-Weierstrass theorem. Convergence of Fourier series is characterized on the unit circle. Wavelets are introduced following the Fourier transform, and their construction as well as ability to approximate functions in L2(R) is discussed. At the end, the universal approximation theorem for artificial neural networks is presented, and the function representation and approximation with single- and multilayer neural networks on R2 is constructed.