An unsupervised learning approach to seismic waveform classification via representation learning
| dc.contributor | Graduate Program in Physics. | |
| dc.contributor.advisor | Özakın, Arkadaş. | |
| dc.contributor.author | Efe, Onur. | |
| dc.date.accessioned | 2025-04-14T12:43:33Z | |
| dc.date.available | 2025-04-14T12:43:33Z | |
| dc.date.issued | 2023 | |
| dc.description.abstract | Seismology is an observation-based science and requires well- classified seismic signals to expand its boundaries. One of the most basic classification problems is to separate earthquake- based seismic activity from background noise. Machine learning algorithms are recently introduced as a solution to this problem. Although unsupervised learning-based algorithms have been used for specific cases (certain seismic events, time intervals, geographical regions, etc.), most proposed algorithms are supervised learning-based. Consequently, general-purpose unsupervised learning algorithms have not been developed yet. Supervised learning-based models may have biases due to the dataset used in training, which can lead to poor results for observational purposes. Considering that the seismic waveforms are labeled in the light of current information, it’s clear that training the models using these labels could block potential discoveries. Developing methods with complementary biases is important to fill the blind spots of supervised learning-based models. Our primary motivation in this thesis is to develop unsupervised learning-based general- purpose detection methods that distinguish earthquake-based seismic activity from background noise. Based on this motivation, deep learning-based methods that can classify waveforms using one or more stations have been developed. | |
| dc.format.pages | xv, 88 leaves | |
| dc.identifier.other | Graduate Program in Physics. MATH 2023 A54 (Thes SCED 2023 I46 | |
| dc.identifier.uri | https://digitalarchive.library.bogazici.edu.tr/handle/123456789/21559 | |
| dc.publisher | Thesis (M.S.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 2023. | |
| dc.subject.lcsh | Seismic waves. | |
| dc.subject.lcsh | Seismology. | |
| dc.subject.lcsh | Supervised learning (Machine learning) | |
| dc.title | An unsupervised learning approach to seismic waveform classification via representation learning |
Files
Original bundle
1 - 1 of 1