Computational Neuroscience Initiative.Behavorial studies have found that humans have the remarkable capability of identifying the basic category, or “gist,” of a scene from just a quick glance of about 100 milliseconds. The goal of my project was to gain insight through machine learning into a way of deducing the “gist” of a scene. The “gist” of something is its most overarching, higher-level meaning. For example, the gist of a scene might be “beach” or “forest.”
Specifically, I examined and compared local textures across images of various scene categories, from cityscapes to forests. These textures are determined by two things: the distribution of light intensities and their spatial organizations relative to each other. Implementing various forms of statistical analyses, from principal component analysis to power spectrum analysis, I probed for some way to differentiate between these different categories of scenes.
Prior to this summer, I had never assisted in research of any sort or even dabbled in theoretical neuroscience. Consequently, starting from scratch, I was able to learn so much about both the research process and the field itself. Through all the guidance of Dr. Balasubramanian and my post-doctorate mentor, Dr. Tesileanu, I got to learn about the technical and mathematical toolbox that theoretical neuroscientists use (which is a lot of linear algebra), specifically relating to image processing.
I also gained understanding of the creative thought processes that go behind research. The process is a series of never-ending questions that sometimes will not necessarily get answered in my lifetime. This inquisitive nature of science has taught me to be more forthright in asking questions and even admitting what I do not know. After all, research is about gleaning new information which cannot be done without establishing what is known and what is not. Being in the presence of such brilliant researchers also helped me grow in various aspects. I was given the opportunity to sit in on various neuroscience research presentations and seminars, so there was never a moment when I was not learning something new.
I would like to sincerely thank everyone who has helped me throughout this project. I would have been thoroughly mired in all the linear algebra and coding bugs had it not been for the mentorship I have been so lucky to receive.
Computational Neuroscience Initiative.Behavorial studies have found that humans have the remarkable capability of identifying the basic category, or “gist,” of a scene from just a quick glance of about 100 milliseconds. The goal of my project was to gain insight through machine learning into a way of deducing the “gist” of a scene. The “gist” of something is its most overarching, higher-level meaning. For example, the gist of a scene might be “beach” or “forest.”
Specifically, I examined and compared local textures across images of various scene categories, from cityscapes to forests. These textures are determined by two things: the distribution of light intensities and their spatial organizations relative to each other. Implementing various forms of statistical analyses, from principal component analysis to power spectrum analysis, I probed for some way to differentiate between these different categories of scenes.
Prior to this summer, I had never assisted in research of any sort or even dabbled in theoretical neuroscience. Consequently, starting from scratch, I was able to learn so much about both the research process and the field itself. Through all the guidance of Dr. Balasubramanian and my post-doctorate mentor, Dr. Tesileanu, I got to learn about the technical and mathematical toolbox that theoretical neuroscientists use (which is a lot of linear algebra), specifically relating to image processing.
I also gained understanding of the creative thought processes that go behind research. The process is a series of never-ending questions that sometimes will not necessarily get answered in my lifetime. This inquisitive nature of science has taught me to be more forthright in asking questions and even admitting what I do not know. After all, research is about gleaning new information which cannot be done without establishing what is known and what is not. Being in the presence of such brilliant researchers also helped me grow in various aspects. I was given the opportunity to sit in on various neuroscience research presentations and seminars, so there was never a moment when I was not learning something new.
I would like to sincerely thank everyone who has helped me throughout this project. I would have been thoroughly mired in all the linear algebra and coding bugs had it not been for the mentorship I have been so lucky to receive.