Sargur N. (Hari) Srihari, 1949-2022
Professor, Scientist, Engineer, and Inventor
Sargur N. (Hari) Srihari, an internationally renowned professor of computer science who taught computers to read handwriting and significantly advanced the fields of pattern recognition, computational forensics, and machine learning, passed away on March 8, 2022. He was 72. A son of India, he came to the United States in 1970 and ultimately settled in Buffalo, NY, serving on the faculty of the University at Buffalo for over four decades.
Professor Srihari established the University at Buffalo as a leading center for pattern recognition and machine learning. He founded the Center of Excellence for Document Analysis and Recognition (CEDAR), which did groundbreaking research work for the U.S. Postal Service in the 1990s, ultimately teaching machines how to read handwritten envelopes. The recipient of seven United States patents, Prof. Srihari’s advances paved the way for handwriting recognition technology used in many modern systems from tablets to scanners. His early research work on three-dimensional imaging also remains influential in fields such as 3-D printing.
He would later become a pioneer in the field of computational forensics. In 2002, he conducted the first computationally-based research work to establish the individuality of handwriting, with important implications for the criminal justice community. He later served on a National Academy of Sciences committee that produced an influential 2009 report on strengthening forensic sciences in the United States. Both his postal-related and computational forensics work received significant media attention. In his later years, he lectured on the topics of machine learning, deep learning, and probabilistic graphical models, with his lecture slides and videos being widely used in courses around the world.
Prof. Srihari was also a highly regarded educator and mentor to his students and colleagues, and a beloved husband, father, and grandfather who lived life to the fullest. Hari is survived by his wife of 45 years, Rohini, by his sons Dileep of Washington, DC and Ashok (Caroline) of Gainesville, FL, by his granddaughter Vera, by his sister Shashi Sampath of Washington, DC, and by his brother Mukund Sargur of Redmond, WA.
Early Life, 1949-1967
Sargur Narasimhamurthy Srihari, known as “Hari” to family, friends, and colleagues, was born in Bangalore, India on May 7, 1949. Following Indian regional tradition, his first name (Sargur) was the name of his family’s ancestral village near Mysore, his middle name was his father’s name, and his surname (Srihari) was his own name. His parents were Prof. S.N. Murthy, a botanist, and his wife Alamelu. His paternal grandfather N. Sama Iyengar (“NSI”) was a district court judge, and had an important influence on young Hari. His maternal grandfather C.L. Iyengar was a civil engineer.
During Hari’s childhood in the early years of Indian independence, his father was a professor of agriculture at Dharwar Agricultural College in northern Karnataka. But due to economic circumstances, the family moved to Mysore for a time where Hari was mentored by his grandfather “NSI” in Sanskrit, Kannada literature, music, and philosophy. They later moved to the Basavangudi neighborhood of Bangalore, eventually to Gandhi Bazaar Road.
In this period, his mother Alamelu took a job at the Life Insurance Corporation of India to provide the family with a source of income, something very unusual for women of her social standing at that time. As a result, his key life goals included keeping a steady job, becoming financially independent, and having the ability to support his own family.
College and University, 1967-1976
Srihari graduated from National College (part of Bangalore University) in 1967 with a B.Sc. in physics and math, equivalent to a Western high school diploma. He obtained his undergraduate degree at the world-renowned Indian Institute of Science in Bangalore, graduating in 1970 with a B.E. degree in Electrical and Communication Engineering. Hari would become a loyal IISc alumnus, returning later in life as a guest lecturer and establishing a scholarship.
For graduate studies, Hari made the fateful decision to immigrate to the United States and came to Ohio State in the fall of 1970. He obtained his M.S. in 1972 and Ph.D. in 1976, both in computer & information science. His doctoral thesis focused on the design and evaluation of classification algorithms for a type of pattern recognition related to radar aircraft identification.
Family and Buffalo, 1976-1991
After graduating from Ohio State, Srihari obtained a faculty position at Wayne State University in Detroit. In the summer of 1976, he met Rohini – the daughter of H.K. Kesavan, a professor and founder of the systems design department at the University of Waterloo, Canada and his wife Rajalaxmi. They were married in March 1977. A year later, Dileep was on the way and Hari and Rohini moved to Buffalo, which had the advantage of being closer to Rohini’s family across the border. Originally intended as a short stay, Hari would eventually build his time at the State University of New York (SUNY) Buffalo into the work of his life.
Srihari’s first major papers were in representation of three-dimensional images, such as those produced in computing tomography. His Association for Computing Machinery (ACM) Computing Surveys article on the topic (1981) was influential, with citations even to this day in 3-D printing. After obtaining tenure in 1982, his work with his first PhD student, on hyper-quad trees for multi-dimensional images, was published in the Communications of the ACM in 1983 and featured on its cover. Meanwhile, Rohini had received her masters’ degree at UB and was teaching at Canisius College during the period when Ashok arrived.
In 1986, the family moved to Meadowview Lane in Amherst, NY which would become their permanent home for the next 35 years. Hari also began pursuing his significant work in pattern recognition. His first external research funding was from the National Science Foundation to study the role of context in recognizing text. Its successful conclusion resulted in a paper in the IEEE Transactions on Pattern Analysis and Machine Intelligence. He then approached the United States Postal Service for funding as there was potential use of his research in machines to read addresses. USPS provided him with a small ($100K) grant to study existing printed address reading technology. This eventually led to large-scale funding from the United States Postal Service for several ancillary projects: determining address blocks on letters, flats and irregular parcels, reading poor quality printed addresses, and ultimately reading handwritten addresses.
CEDAR and Professional Acclaim, 1991-2001
In 1991, the USPS recognized Srihari’s UB laboratory as a Center of Excellence for Document Analysis and Recognition, known as CEDAR. Under his direction, the center’s work eventually involved more than a dozen projects and at its peak, more than a hundred graduate students and a dozen full-time research staff. Total funding for the center exceeded $60 million over 25 years and Srihari also obtained six United States patents related to the work. Meanwhile, as Dileep and Ashok were growing up, Hari’s support helped Rohini to finish her PhD, after which she also joined the UB faculty.
Due to Hari’s work at CEDAR in the 1990s, handwritten digit recognition was recognized as the “fruit fly” of AI and machine learning. Meanwhile, document analysis was recognized as an important application area of pattern recognition and machine learning, spawning several conferences and a journal. The first large-scale handwritten address interpretation (HWAI) system in the world was deployed by the USPS, eventually reading over 90% of handwritten US mail and spraying barcodes on the envelopes. This paved the way for the handwriting recognition technology now found in every tablet device. This work is exhibited in the National Postal Museum in Washington, DC.
The work of CEDAR during this period received worldwide attention. Srihari was featured on ABC News, the Discovery Channel, and in articles in the New York Times and other major publications. Professional recognition followed as well, as he was named a fellow of the Institute of Electrical and Electronics Engineers (IEEE) in 1994 and a fellow of the International Association for Pattern Recognition in 1996. In 1997, he was made a State University of New York (SUNY) Distinguished Professor, the university’s highest academic faculty rank. He also served as general chair or co-chair of the International Conference on Document Analysis and Recognition (ICDAR) and the International Workshop on Frontiers in Handwriting Recognition (IWFHR) on multiple occasions.
Computational Forensics and Teaching, 2001-2015
During the early 2000s, Hari increasingly spent time supporting Rohini in her dual roles as a UB faculty member and entrepreneur who started multiple businesses. Meanwhile, for several decades, Hari’s research career involved regular trips to Washington, DC to visit various federal funding agencies.
Around the same time, Hari began turning his research focus to forensic science. The National Institute of Justice (NIJ) contacted him about the need for a scientific basis for continuing to allow impression evidence such as handwriting to be introduced as evidence in court. His first effort was on quantifying the individuality of handwriting. A resulting 2002 paper in the Journal of Forensic Sciences was hailed by the community as providing a basis for admitting handwriting evidence in several cases. Srihari himself testified in several “Daubert hearings” as to whether handwriting could be admitted as evidence. A seventh patent related to this work was obtained in 2009.
This handwriting work led to the first automated system, known as CEDAR-FOX, for determining whether two handwritten samples came from the same or different writer. The handwriting work was eventually extended to comparing fingerprints and footwear prints. As a result, he was invited to serve as the only computer scientist on the National Academy of Sciences’ Committee on Identifying the Needs of the Forensic Science Community. The committee’s landmark 2009 report, Strengthening Forensic Science in the United States: A Path Forward, has had a major impact in courts worldwide and received an award from the Innocence Project. Hari and CEDAR-FOX were later featured in a 2013 NOVA episode on PBS regarding handwriting evidence related to the Lindbergh kidnapping.
Later Years, 2015-2022
Eschewing retirement, Hari continued actively teaching and supervising graduate students. He ultimately taught courses in artificial intelligence, pattern recognition and machine learning for over four decades. He also supervised 43 doctoral students to completion along with hundreds of masters’ students, and in 2018 received UB’s Excellence in Graduate Mentoring Award. He also developed an extensive set of lecture slides covering three courses: Introductory Machine Learning, Deep Learning, and Probabilistic Graphical Models. His lecture slides and videos are widely used in courses around the world.
His final teaching efforts were focused on integrating the avalanche of research being produced in deep learning from various books, papers and blogs. He returned to India during the spring of 2020 to spend time with his mother and serve as a Visiting Professor / Scientist at his alma mater, the Indian Institute of Science, and later established a scholarship there. During the COVID-19 pandemic, he began recording videos of his explanation of topics in deep learning, and did livestreaming as well. His lectures on topics such as attention models received many views online.
In his later years, Hari also continued enjoying traveling to Washington with Rohini to enjoy Dileep’s music concerts, along with trips to Florida to visit Ashok and Caroline. He continued to read avidly while pursuing his lifelong love of history, science, and gardening. He also continued supporting his immediate and extended family in important ways. He was delighted to welcome his granddaughter Vera in December 2020, and she was the apple of her grandfather’s eyes. He passed away on March 8, 2022 in Washington, DC at the age of 72 in the presence of his family, due to complications from a glioblastoma. Hari will be missed dearly by all who knew him.
S.N. Srihari is survived by his wife of 45 years, Rohini, and by his sons Dileep Srihari of Washington, DC and Ashok Srihari (Caroline) of Gainesville, FL. He is also survived by his granddaughter Vera Srihari.
He is further survived by his sister Shashi Sampath of Washington, DC, by his brother Mukund Sargur of Redmond, WA, by Rohini’s mother Rajalaxmi and her sisters Anita and Kalpana.
Moving Scribbled Mail Along, New York Times, May 27, 1992
Digitally Detecting Forgeries, WIRED, June 1, 1999
Prosecutors Hope New Study of Handwriting Analysis Will Silence Skeptics, New York Times, May 26, 2002
How Do Post Office Machines Read Addresses?, LiveScience.com, August 20, 2007
Artificial Examiners Put to the Test, BBC, August 24, 2007
Plugging Holes in the Science of Forensics, New York Times, May 11, 2009
Groundbreaking computer scientist Sargur Srihari dies at 72, University at Buffalo, March 10, 2022
Selected Video Reports
ABC World News Tonight with Peter Jennings, June 18, 1997
Discovery Channel, July 1997
MCTV News, February 2003
Who Killed Lindbergh’s Baby?, NOVA (PBS), season 40 ep. 4, January 2013 (jump to 32:04)