I would appreciate any feedback regarding any aspect of my SoP.
Prompt
This document should explain your rationale for obtaining a graduate degree in Computer Science -- your research interests, what you bring to the program, and how you hope to use the degree. This is not a "life story", but be sure to cover any aspects of your application/background that might need additional explanation. Remember that readers of your statement are CS researchers, but not necessarily in your area!
Ever since I was a kid, I found the challenge of analyzing and solving math problems or riddles very fulfilling. Eventually, I realized that I could put this passion to use by creating computer programs and algorithms that can have an impact on the real world, which is what led me to pursue this area of study. During my freshman year, I came upon an application that could seemingly understand what an image was depicting. Immediately, I became curious about the algorithm that performed such a complex task. Even though image classification was the first, I was introduced to many more applications that were driven by machine learning and data during my studies. This variety of ground-breaking applications, which were so deeply rooted in the real world due to the data they use, solidified my view of this area of computer science as something special whose potential was limitless. These factors along with my desire to possess a deep understanding of my discipline led me to direct my efforts towards gathering theoretical and practical experience in various machine learning methodologies through university and online courses, projects, and work experience with the aim of one day joining the innovators of the tech industry. I wish to delve even deeper into these topics, and I believe your master's program will allow me to do that by giving me the opportunity to cooperate with and learn from leading scientists and engineers.
I had an enriching learning experience throughout my bachelor studies at NTUA. I built strong theoretical and practical foundations on the principles of machine learning, software engineering, statistics, and data structures. Through challenging assignments and projects, I understood the importance of being methodical and rigorous in the way I work and present my ideas, a practice that is reflected throughout my studies and work. Having a strong desire to explore and understand machine learning and data science, I attended courses such as "Neural Networks", "Computer Vision", "Artificial Intelligence" and "Pattern Recognition", some of them in earlier semesters than those offered. Even in courses that were not directly relevant, I sought to integrate algorithms that learn from data. One notable research project that was done as part of the course "Electronic Health Technologies" was the development of a chatbot with mood-tracking capabilities. I reviewed existing methods for the classification of text to emotion and the theory behind emotion modeling in continuous space to develop an application that consisted of a neural network that analyzed the user's responses and categorized them into one of five basic emotions. My ongoing thesis titled "Discovering interpretable directions in the latent space of neural networks guided by natural language" involves creating a framework that allows the automatic detection of interpretable directions in the latent space of invertible generative models such as Glow with the help of text-assisted image generation. Through this endeavor, I have become accustomed to the nuances and challenges of research.
Seeking to gain practical experience by working in an industry setting, I joined u-blox as an intern during my senior year, working on the development of an indoor localization system from scratch, in collaboration with Libra AI Technologies. I decided to grasp the opportunity and continue my work on this project after the end of the internship by joining Libra, prioritizing it over my thesis. I proposed an architecture that makes use of multiple sensors in a single predictor to deal with situations where the transmitter was not in the line of sight of a sensor, resulting in bad signal quality. This work was published as "Deep Learning-Based Indoor Localization Using Multi-View BLE Signal" in Sensors Journal and has given me essential insights on the importance of designing and conducting tests to evaluate various aspects of the system such as robustness to slight changes in the position of the furniture and sensors. In Libra, I also worked on the development of a keyword extraction algorithm. The algorithm operates on data from Twitter which has many undesirable features such as irregular language and sentence length. The final system, which is now an integral part of the client's application, consists of a semantic search module that retrieves relevant tweets, a pre-processing pipeline, a syntax-based keyword extraction module, and a scoring function that is based on embedding similarity. Through this project, I became proficient in many natural language processing methodologies and gained practical experience with transformer models. I have now returned to university to complete my thesis and I am excited to return to Libra the coming March.
I aspire to bridge the gap between industry and research through my work by bring recent developments in the areas of natural language processing and machine learning in the real world. To be more specific, I am interested in areas such as text summarization, machine translation and question answering. I am also a dedicated language learner and aspire to enhance language exchange opportunities through my work. I believe that professors such as Greg Durett and Eunsol Choi complement my interests, and it would be an honor to work with them. I am sure that the stimulating academic environment and interaction with the distinguished faculty will prove immensely fruitful and facilitate my growth both as an innovator and as a person. For these reasons I would consider it a privilege to be able to attend your master's program.
Prompt
This document should explain your rationale for obtaining a graduate degree in Computer Science -- your research interests, what you bring to the program, and how you hope to use the degree. This is not a "life story", but be sure to cover any aspects of your application/background that might need additional explanation. Remember that readers of your statement are CS researchers, but not necessarily in your area!
Statement of Purpose
Ever since I was a kid, I found the challenge of analyzing and solving math problems or riddles very fulfilling. Eventually, I realized that I could put this passion to use by creating computer programs and algorithms that can have an impact on the real world, which is what led me to pursue this area of study. During my freshman year, I came upon an application that could seemingly understand what an image was depicting. Immediately, I became curious about the algorithm that performed such a complex task. Even though image classification was the first, I was introduced to many more applications that were driven by machine learning and data during my studies. This variety of ground-breaking applications, which were so deeply rooted in the real world due to the data they use, solidified my view of this area of computer science as something special whose potential was limitless. These factors along with my desire to possess a deep understanding of my discipline led me to direct my efforts towards gathering theoretical and practical experience in various machine learning methodologies through university and online courses, projects, and work experience with the aim of one day joining the innovators of the tech industry. I wish to delve even deeper into these topics, and I believe your master's program will allow me to do that by giving me the opportunity to cooperate with and learn from leading scientists and engineers.
I had an enriching learning experience throughout my bachelor studies at NTUA. I built strong theoretical and practical foundations on the principles of machine learning, software engineering, statistics, and data structures. Through challenging assignments and projects, I understood the importance of being methodical and rigorous in the way I work and present my ideas, a practice that is reflected throughout my studies and work. Having a strong desire to explore and understand machine learning and data science, I attended courses such as "Neural Networks", "Computer Vision", "Artificial Intelligence" and "Pattern Recognition", some of them in earlier semesters than those offered. Even in courses that were not directly relevant, I sought to integrate algorithms that learn from data. One notable research project that was done as part of the course "Electronic Health Technologies" was the development of a chatbot with mood-tracking capabilities. I reviewed existing methods for the classification of text to emotion and the theory behind emotion modeling in continuous space to develop an application that consisted of a neural network that analyzed the user's responses and categorized them into one of five basic emotions. My ongoing thesis titled "Discovering interpretable directions in the latent space of neural networks guided by natural language" involves creating a framework that allows the automatic detection of interpretable directions in the latent space of invertible generative models such as Glow with the help of text-assisted image generation. Through this endeavor, I have become accustomed to the nuances and challenges of research.
Seeking to gain practical experience by working in an industry setting, I joined u-blox as an intern during my senior year, working on the development of an indoor localization system from scratch, in collaboration with Libra AI Technologies. I decided to grasp the opportunity and continue my work on this project after the end of the internship by joining Libra, prioritizing it over my thesis. I proposed an architecture that makes use of multiple sensors in a single predictor to deal with situations where the transmitter was not in the line of sight of a sensor, resulting in bad signal quality. This work was published as "Deep Learning-Based Indoor Localization Using Multi-View BLE Signal" in Sensors Journal and has given me essential insights on the importance of designing and conducting tests to evaluate various aspects of the system such as robustness to slight changes in the position of the furniture and sensors. In Libra, I also worked on the development of a keyword extraction algorithm. The algorithm operates on data from Twitter which has many undesirable features such as irregular language and sentence length. The final system, which is now an integral part of the client's application, consists of a semantic search module that retrieves relevant tweets, a pre-processing pipeline, a syntax-based keyword extraction module, and a scoring function that is based on embedding similarity. Through this project, I became proficient in many natural language processing methodologies and gained practical experience with transformer models. I have now returned to university to complete my thesis and I am excited to return to Libra the coming March.
I aspire to bridge the gap between industry and research through my work by bring recent developments in the areas of natural language processing and machine learning in the real world. To be more specific, I am interested in areas such as text summarization, machine translation and question answering. I am also a dedicated language learner and aspire to enhance language exchange opportunities through my work. I believe that professors such as Greg Durett and Eunsol Choi complement my interests, and it would be an honor to work with them. I am sure that the stimulating academic environment and interaction with the distinguished faculty will prove immensely fruitful and facilitate my growth both as an innovator and as a person. For these reasons I would consider it a privilege to be able to attend your master's program.
