Sebastian Seung

From Wikipedia, the free encyclopedia
Jump to navigation Jump to search
H. Sebastian Seung
CitizenshipAmerican
EducationHarvard University (PhD)
Hebrew University of Jerusalem (PostDoc)
Known forConnectome Theory
Non-negative matrix factorization
Children3 daughters
AwardsSloan Fellowship
Scientific career
Fieldsneuroscience
physics
InstitutionsPrinceton University
MIT
Bell Labs
ThesisPhysics of Lines and Surfaces (1990)
Doctoral advisorDavid Robert Nelson
Korean name
Hangul
Hanja
Revised RomanizationSeung Hyeonjun
Websitehttp://seunglab.org/

Sebastian Hyunjune Seung (English: /sung/ or [səŋ]; Korean승현준; Hanja承現峻)[1][2] is a Korean-American multi-disciplinary scientist whose research efforts have spanned the fields of neuroscience, physics, computer science, bioinformatics, machine learning, deep learning, and artificial intelligence.[3] He has helped pioneer the new field of connectomics, "developing new computational technologies for mapping the connections between neurons," and has been described as the cartographer of the brain.[4][5]

Since 2014, he has been a professor in computer science and neuroscience at Princeton University's Neuroscience Institute at the Jeff Bezos Center in Neural Dynamics, where he directs the Seung Labs.[6] Before, he worked at the Massachusetts Institute of Technology as a full professor in computational neuroscience in the Department of Brain and Cognitive Sciences and as a professor in physics.

In the industry, he was a research scientist at the Bell Labs and an Investigator of the Howard Hughes Medical Institute.[7] Since 2015, he has joined the board of advisors for Nara Logics, an MIT-based startup specializing in brain research and big data.[8] Since 2018, he was hired as the Chief Research Scientist at Samsung.[9][10]

He is most well known as a proponent of connectomics through his Ted talk "I am my Connectome" and his book Connectome which was named top 10 nonfiction books of the year 2012 by the Wall Street Journal and has been translated into dozens of languages.[11]

He has also founded EyeWire, an online computer game that mobilizes social computing and machine learning on a mission to map the human brain. It has attracted hundreds of thousands of users from over a hundred countries, and it has recently partnered with KT Corporation to help spread the scientific mission and attract more players to the cause.

Seung is also known for his 1999 joint work on non-negative matrix factorization, an important algorithm used in AI and data science.[12]

Biography[edit]

Seung was born in 1966 in Austin, Texas. His father Thomas Seung is a philosophy professor at the University of Texas, Austin, and Korean-American immigrant who escaped North Korea as a teenager.[5] Sebastian's mother is Kwihwan Hahn, a graduate of Julliard, and he has two younger siblings, a brother, currently a professor at Harvard Medical School, and a sister, currently a psychiatrist.

By age five, he had taught himself how to read. Growing up, his passions were soccer, math, nonfiction (science and philosophy), and Greek myths. His interest in western philosophy and the classics appears in his books including Connectome. As a teenager, he was particularly inspired by Carl Sagan's "Cosmos" to become a physicist.[5]

Education and physics career (1982-2005)[edit]

He studied theoretical physics as an undergraduate at Harvard University (enrolled 1982 when 16 years old), taking graduate courses as a sophomore when he was 17 years old.[5] He then went straight into Harvard's graduate program and obtained his Ph.D. in 1990 under the supervision of David Robert Nelson.

Seung's 1990 doctoral dissertation is titled "Physics of Lines and Surfaces." It examines the statistical mechanics of vortex lines in high-temperature superconductors and uses tools such as the renormalization group perturbation theory. It then uses Monte Carlo simulations to analyze buckling phase transition behavior and critical phenomena, drawing comparisons with the Ising model and XY spin-glass model. Finally it introduces a continuum elastic theory for certain hexatic molecules.[13][14]

During his Ph.D studies he briefly interned at the Bell Labs in 1989. There he was introduced to the mathematical problem of neural networks.[5]

He completed his postdoctoral training at the Hebrew University of Jerusalem. He returned to the Bell Labs and was a member of the Theoretical Physics Department.[5] In 2004, he joined the MIT faculty first as a professor in physics and then as a professor in neuroscience.

Switch to neuroscience and connectomics[edit]

It was near the end of 2005 when he made the switch from physics to neuroscience, which at the time was considered a risky career move. In November, one of his former mentors David Tank from the Bell Labs suggested a new problem to Seung: how does the brain work? He was invited to a neuroscience conference in Germany, and in January 2006 he brought two of his graduate students to learn about a new technology that imaged the brain in higher resolution built by Winfried Denk. It was then that Seung worked day and night writing grant proposals to fund computational research in connectomics, which at the time was seen as a "highly speculative engineering project." [5]

Since 2014, Seung joined the faculty at Princeton as a professor in neuroscience at the Bezos Center for Neural Circuit Dynamics. Seung now leads a team working on an online citizen science project, EyeWire. It is human-based computation game about tracing neurons in the retina. The game was developed by MIT and the Max Planck Institute for Medical Research.[15]

The Connectome Theory[edit]

The connectome is the map of the 100 trillion plus neural connections within the brain. Its name is based on the same way the genome is a map of a species' DNA. In simplest mathematical terms, it can be thought of as a graph network. Seung focuses on the potential implications of the Human Connectome Project and what it would mean to map the connectome of a human brain. He has popularized the connectome theory through his 2010 TED Conference speech titled I Am My Connectomeas well as through his 2012 book Connectome: How the Brain’s Wiring Makes Us Who We Are.[16]

He proposes that every memory, skill, and passion is encoded somehow in the connectome. And when the brain is not wired properly it can result in mental disorders such as autism, schizophrenia, Alzheimer's, and Parkinson's. Understanding the human connectome may not only help cure such diseases with treatments but also possibly help doctors prevent them from occurring in the first place. And if we can represent the sum of all human experiences and memories in the connectome, then we can download human brains on to flash drives, save them indefinitely, and replay those memories in the future, thereby granting humans a kind of immortality.[5]

TED Talk: "I Am My Connectome" [edit]

In his 2010 TED Conference speech, Seung hypothesizes that the essence of a human being is his or her connectome. The complexities and vast amount of neural connections in the human brain has slowed the complete mapping of the human connectome. This is in comparison to the only completely mapped connectome to date, that of the nematode Caenorhabditis elegans, a process that took over 12 years to complete despite the animal's hermaphrodite form only having a total of 302 neurons in its entire nervous system.[17]

Seung proposes that a connectome is like a riverbed. As the water of a river, neural activity is constantly changing, never staying still. The connectome is the riverbed which both guides the neural activity while also being shaped by the water over time. Illustrating how thinking and neural activity alters the connectome adding to the difficulty of mapping the human connectome that is constantly changing.

Connectome: How the Brain’s Wiring Makes Us Who We Are [edit]

In his 2012 book Connectome, Seung discusses his current views on neuroscience and the upcoming science of connectomics. The book expands on some of the concepts discussed in his Ted talk as well as discussing how the doctrine of the connectome can be tested. He states that in order to test and further our knowledge and unlock to potential of the connectome we must improve the scientific tools in existence. Also, he states that there needs to be new ways to promote the concept of the connectome using the four R's: reweighting, reconnection, rewiring, and regeneration.[16]

EyeWire.org[edit]

EyeWire is a computer game developed by Seung designed to map neuron cells in the human brain. Users can sign up for free, and the game helps contribute to ongoing cutting-edge scientific research. In Seung's own words:

We have this new site: Eyewire.org. It is a citizen science project. Our AI is not accurate enough to map the connectome by itself. We still need human intervention. So we have now created this website that allows anybody to do it.[18]

Thusfar site has recruited over 130,000 players from over 100 countries.[4] KT Corporation, South Korea's largest telecom provider, recently partnered with EyeWire to advertise the game across the country and attract more players.

Essentially, in the game one has to identify and color connected components of neuron cells just from the 2d cross sections of brain tissue. As explained in his book "Connectome," up to now neuroscientists can only accurately image brain tissue using 2d sections (as opposed to 3d scans), which necessitates the need to splice these 2d pictures together to create a neural network map of the brain's inner connections. While artificial intelligence and computer vision can perform some of the manual work, it still takes a combined computer-human effort to map something as huge as the human brain, a computational endeavor that has perhaps never been attempted before at this scale in human history, hence the need for social computing.

Ongoing experiments[edit]

In the same way the Human Genome Project and the complete mapping of human DNA has helped reveal a lot about human biology, Seung and other connectomists hope that a complete map of the human brain can reveal a lot about how we humans think and perceive, how memory works, important questions that has been asked since the time of Aristotle, and with connectomics we could be on the verge of answering them scientifically.

A team at Janelia plans to map the connectome of Drosophila by around 2025.[19] Seung also helped set up experiments with Tank and Nobel Laureate Richard Axel to find memories in the connectome.[5]

Publications and books[edit]

His algorithms for nonnegative matrix factorization have been widely applied to problems in visual learning, semantic analysis, spectroscopy, and bioinformatics. He continues to study neural networks using mathematical models, computer algorithms, and circuits of biological neurons in vitro.

As aforementioned he authored the book Connectome (2012). It has been translated into at least 26 languages.

He has published many other scholarly papers. A selection is published on his website:

https://pni.princeton.edu/faculty/h.-sebastian-seung

Awards and honors[edit]

He has been a Sloan Research Fellow, a Packard Fellow, and a McKnight Scholar. He has also won the Ho-am Prize in Engineering and has been named top 10 non-fiction authors by the WSJ for his book Connectome.[20] He is an External Member for the Max Planck Society.[4]

Teaching[edit]

"He is a popular teacher who traveled the world—Zurich; Seoul, South Korea; Palo Alto, California—delivering lectures on his mathematical theories of how neurons might be wired together to form the engines of thought."[5]

In the past few years, he's been teaching Princeton's COS 485 Neural Networks, a course taken by both undergraduates and graduate students.

Personal life[edit]

He currently lives with his wife and 3 daughters. Their first was born in January 2006, as he was making his career switch from physics to neuroscience.[5]

He was known to be "so naturally exuberant that he was known for staging ad hoc dance performances with Harvard Square's street musicians." He continues to enjoy playing soccer in the fields of Princeton. And he enjoys eating mixed nuts from Costco.[5]

See also[edit]

References[edit]

  1. ^ chosun.com (Korean)
  2. ^ "삼성전자, AI 분야 세계적 석학 '세바스찬 승·다니엘 리' 영입". biz.chosun.com (in Korean). Retrieved 2019-04-10.
  3. ^ "H. Sebastian Seung | Neuroscience". pni.princeton.edu. Retrieved 2019-04-11.
  4. ^ a b c "Sebastian Seung". edX. 2014-08-12. Retrieved 2019-04-14.
  5. ^ a b c d e f g h i j k l Cook, Gareth (2015-01-08). "Sebastian Seung's Quest to Map the Human Brain". The New York Times. ISSN 0362-4331. Retrieved 2019-04-14.
  6. ^ http://seunglab.org
  7. ^ http://www.hhmi.org/scientists/h-sebastian-seung
  8. ^ "Xconomy: Nara Logics Gets New CEO, Recruits Renowned Neuroscientist". Xconomy. 2015-05-05. Retrieved 2019-04-11.
  9. ^ "World-Renowned AI Scientists, Dr. Sebastian Seung and Dr. Daniel Lee Join Samsung Research". news.samsung.com. Retrieved 2019-04-10.
  10. ^ "Samsung hires two world-renowned AI experts". english.donga.com. Retrieved 2019-04-10.
  11. ^ "The Best Nonfiction of 2012". Wall Street Journal. 2012-12-14. ISSN 0099-9660. Retrieved 2019-04-10.
  12. ^ H. Sebastian Seung; Lee, Daniel D. (October 1999). "Learning the parts of objects by non-negative matrix factorization". Nature. 401 (6755): 788–791. doi:10.1038/44565. ISSN 1476-4687. PMID 10548103.
  13. ^ Mathematics Genealogy Project
  14. ^ Seung, Hyunjune Sebastian (1990-01-01). "Physics of Lines and Surfaces". Ph.D. Thesis. Bibcode:1990PhDT.......194S.
  15. ^ "About << EyeWire". Retrieved March 26, 2012.
  16. ^ a b Seung, Sebastian (2012). Connectome: How the Brain's Wiring Makes Us Who We Are (1 ed.). New York, New York: Houghton Mifflin Harcourt Publishing Company. ISBN 9780547678597.
  17. ^ White, J. G.; Southgate, E.; Thomson, J. N.; Brenner, S. (1986). "The Structure of the Nervous System of the Nematode Caenorhabditis elegans". Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences. 314 (1165): 1–340. doi:10.1098/rstb.1986.0056. JSTOR 2990196.
  18. ^ Ouellette, Jennifer. "The Age of Connectome: Q&A with Sebastian Seung". Scientific American Blog Network. Retrieved 2019-04-10.
  19. ^ "FlyEM | Janelia Research Campus". www.janelia.org. Retrieved 2019-04-14.
  20. ^ "Bio". CIFAR. Retrieved 2019-04-10.

Other references[edit]