- I am know that the program will be a challenging and require me to grow beyond my comfort zone.
My experiences of overcoming difficulties in academics, industry, and self-study makes me well-prepared for this.
In academdia,
Applied Mathematics and Computer Science master's degree gives me a strong background for applying logical and theoretical concepts.
- Courses such as Linear Algebra, Algebra, Analysis, Operations Research provide a solid background to understand ideas at a fundamental level.
- In master's thesis, analyzing protein structure using CNNs and persistant homology, gave me valuable interdisciplinary insight on how math and CS can compliment each other.
- Futhermore, I found myself enjoying the many hours thinking, meditating, and exploring theoretical concepts and questions.
In the industry,
Working in the industry gave me valuable experience overcoming difficult problems and producing consistant results.
- Here are three examples.
- First, working on segmentation of satellite images gave me opportunity to read, implement from scratch, and benchmarked papers such as U-Net and Deep Aggregate Net.
- The prototype and analysis data was used to win a major government contract for our company.
- Second, while developing a semiconductor-robot monitoring prototype for SK hynix, I combined Novel View Synthesis CNN algorithm with pose estimation to track and pose the robot arm.
- The prototype was used to automatically extract to extract key details from a complex semiconductor monitoring video: the position of the robotic arm, the various nozzles attached to the arm, when the nozzle was spraying a solution or not
- Finally, predicting 3D structure from 1D protein sequences gave me experience of managing and finishing a complex project.
- Filter and preprocess large datasets of 25,000 proteins in various formats with missing information.
- Combined various neural network architectures such as, LSTM, CNNs, and ESM language models embeddings.
- Research and experiment with distance, dihedral angle, frame-aligned point error functions.
- Much more stable, consistant, and quickly converging model from the starting baseline, Recurrent Geometric Networks.
- In these three cases I learned gained common insights that helped me break through challenges:
- have cautious optimism that a nice-simple solution exists, focus on key objectives, and work at a steady-consistant pace
In my self-studies, pursuing challenging independent research interests taught me discipline, self motivation and drive.
- Studying the AlphaGoZero paper, I implemented a Monte Carlo Search Tree model-based RL to play variations of tic-tac-toe on longer length-to-win and boards.
- It was very challenging to get the initial prototypes to work on larger variations of the game; I carefully and patiently tried many different ideas.
- Eventually, reasonable success was found using: multi-step lookahead using model value estimation, augmentation of board inputs using rigid transformations, adjustments on rollout sample size depending on game depth, and many minor changes.
- Wanting to underestand the RL the much better, I am studying the fundamentals of studying markov decision process programming and optimal control.