Universal Coding of Memoryless Sources over large (possibly infinite) alphabets
Event details
| Date | 26.06.2024 |
| Hour | 10:00 › 12:00 |
| Speaker | Emre Serhat Coban |
| Location | |
| Category | Conferences - Seminars |
EDIC candidacy exam
Exam president: Prof. Michael Gastpar
Thesis advisor: Prof. Yanina Shkel
Thesis coadvisor: Prof. Emre Telatar
Co-examiner: Prof. Nicolas Macris
Abstract
Source Coding (or Compression) is one of the most fundamental problems in Information Theory. When the distribution of the source alphabet is known, Huffman coding achieves the entropy of the source alphabet (for finite alphabets) which is the fundamental limit. However, in most of the real life scenarios, the source alphabet is not known. In this case, it is not possible to have zero redundancy in one-shot. However, there are several Universal Compression algorithms for which the redundancy diminishes in the asymptotical case. In this work, I studied the background of Universal Compression along with the more recent Universal Coding techniques for any alphabet, including infinite and even unknown alphabets, that achieves diminishing redundancy. Moreover, I examined the performance analysis of some approaches that are specific to the large, possibly infinite, source alphabets.
Background papers
coming soon
Exam president: Prof. Michael Gastpar
Thesis advisor: Prof. Yanina Shkel
Thesis coadvisor: Prof. Emre Telatar
Co-examiner: Prof. Nicolas Macris
Abstract
Source Coding (or Compression) is one of the most fundamental problems in Information Theory. When the distribution of the source alphabet is known, Huffman coding achieves the entropy of the source alphabet (for finite alphabets) which is the fundamental limit. However, in most of the real life scenarios, the source alphabet is not known. In this case, it is not possible to have zero redundancy in one-shot. However, there are several Universal Compression algorithms for which the redundancy diminishes in the asymptotical case. In this work, I studied the background of Universal Compression along with the more recent Universal Coding techniques for any alphabet, including infinite and even unknown alphabets, that achieves diminishing redundancy. Moreover, I examined the performance analysis of some approaches that are specific to the large, possibly infinite, source alphabets.
Background papers
coming soon
Practical information
- General public
- Free