Meeting #4005
Updated by Evgeniy Pavlovskiy almost 5 years ago
h1. 1. Enterprize Practice from "Digital Digital Field Technologies":http://digitalfield.ru/ Technologies _by Dmitry Tailakov_ Part-time, flexible schedule. Ability to write theses and participation in real projects. Recommended requirements: - knowledge of language python|C++; - the ability to use the library for data analysis Pandas, NumPy, SciPy, Matplotlib, TensorFlow, Keras; - algorithm using optimization techniques and machine learning methods; - possession of numerical methods of mathematical modeling in hydrodynamics and geophysics. Contact person: _Dmitry Tailakov_, CTO Digital Field Technologies: dtailakov@nntc.pro h1. 2. "Topics":/attachments/download/3197 for master thesis from the Digital Field Technologies company Topics are attached to this issue. Contact person: _Dmitry Tailakov_, dtailakov@nntc.pro h1. 3. "SDAML lab.":https://bigdata.nsu.ru/lab/ Enterprize Practice. Universal Practice Plan (180h, 1/3 part-time - 10-20k RUB) : 1. A review of modern solutions based on technical requirements 2. Define metrics for solving the problem (WER, MOS, ...) 3. Find dataset, access (LibriSpeech, VoxCeleb) 4. Find Implementation 5. Reproduce baseline 6. Wrap into a package/module/web-service (define and implement interfaces) 7. Document results 8. [Add data to the dataset] 9. [Retrain with new dataset] 10. Present results Date start: 5-Oct Test period - 1 month. Topics (6 students - max): 1. Voice Command Recognition 2. Speaker Identification 3. Speech-to-Text 4. Text-to-Speech 5. Tumor classification 6. Bones cut Contact person: _Evgeniy Pavlovskiy_, leading researcher SDAML lab. NSU: e.pavlovskiy@g.nsu.ru. h1. 4. Papers assignment Current assignment: !image_2020-09-22_17-23-37.png! Decided to form a list of reproducible papers with the help of *Alix Bernard* - by tomorrow (25.09.2020). Then vote for reproducible papers to select it for inassigned 2nd year students.