Postdocs Power AI Research

Postdoctoral researchers play a special role in academic research, bringing in new ideas and energy while they advance their careers into independent scholars. Stony Brook’s Institute for AI-Driven Discovery and Innovation is benefiting from two new postdoctoral associates, Baojian Zhou and Naoya Inoue, both of whom were welcomed to the team this past year. They are spearheading new research which will advance our understanding of online optimization, explainable AI, natural language processing, and machine learning. By working with diverse faculty and graduate students across many research groups, their presence amplifies activities across the Institute.

Their stories are revealing. Baojian Zhou’s path started with his education at Anhui University in China, now the home of a . Zhou’s interest was piqued when he encountered combinatorial problems from discrete mathematics and graph theory in his undergraduate algorithms course. After receiving his bachelor's degree from Anhui University and his master’s degree from Beihang University, Zhou made the concerted effort to earn a PhD in the United States. He landed at the University of Albany, where he concentrated his studies in data mining and machine learning to continue his exploration of mathematical models. Zhou graduated with his PhD in computer science and a master's degree in mathematics early in 2020.

A chance meeting with a Stony Brook graduate student at the ACM Knowledge and Data Discovery Conference (KDD) tipped Zhou off about postdoctoral positions at the Institute. Zhou went on to meet Steve Skiena, Director of Stony Brook’s AI Institute, and they bonded over shared research interests in data mining, data science, and machine learning . “I discovered that Stony Brook is the perfect place to work with diligent and inspired faculty, and to expand my research experience,” says Zhou.

Baojian Zhou has dedicated his Stony Brook research predominantly toward graphical representation learning, with the goal to design models that can comprehend network data. Coupled with his focus in online optimization, these two domains create a platform to amplify knowledge of user behavior in social media.

“Through embarking on this new research, we can garner a better understanding of social media users’ behavioral patterns,” says Zhou. In a society dominated with social networking, this behavioral understanding lends itself to its impact on a corporate, academic, and personal level.

Since arriving at Stony Brook, he has been working with several faculty members at the university, including Yifan Sun, assistant professor of Computer Science. Professor Sun, an expert in online optimization, has been working closely with Zhou to further his progress in optimization to deal with graphical data. Zhou credits both Professors Skiena and Sun as being cornerstones of his research.

The second postdoctoral scholar, Naoya Inoue, left a faculty position in Japan to join the Institute. He was inspired by artificial intelligence in childhood and blossomed with his introduction to programming,

“When I was 10 years old my father was trying to learn programming, and following in his footsteps, I hand-typed a sample game program there into the computer. From then on I fell in love with programming,” says Inoue.

In his home country of Japan, Inoue received his bachelor’s degree in economics. Although this field strays from artificial intelligence, “I enjoyed seeing how human behaviours are modeled in a mathematical manner,” says Inoue. After graduating, Inoue was introduced to Natural Language Processing and went on to receive a master’s degree in the field from Nara Institute of Science and Technology in 2010. Just three years later, he earned his PhD in information science from Tohoku University. But Inoue wanted to gain research experience in the United States.

The value of working with those from diverse backgrounds is indispensable. Inoue says that “the AI Institute at Stony Brook is a great choice for me because it has researchers and students from diverse backgrounds, and its research vision matched with my vision of AI.” While Stony Brook was enticing, Inoue’s decision was not so simple. With his family and wife back in Yamanashi, Japan, leaving them was a difficult decision, particularly in a pandemic. However, with the unwavering support from his family and the Institute, Inoue rests assured that he made the best decision as he embarks on his new research endeavors at Stony Brook.

Naoya Inoue has set his research focus toward machine learning applied to natural language processing and machine reading comprehension. He has tried to tackle disparities within machine intelligence to develop machines that can think and explain their own thinking.

“Although some studies report that computers already have a reading comprehension ability almost equivalent to that of humans,” says Inoue. “The truth is that machines are not so strong, because they cheat.”

Inoue has focused his research towards quantifying machines’ true level of reading comprehension. This extends itself to “explainable AI,” which is artificial intelligence that can explain its own thinking. Inoue’s work with AI faculty members and students has greatly advanced this work. With respect to his research in developing quantitative data, Inoue has been working closely with Niranjan Balasubramanian, assistant professor of the Department of Computer Science, and several graduate students. The group meets weekly to discuss their research progress. He is also working with Steven Skiena and his students to advance their research with explainable AI.

“All of these collaborations will make for better research, and will make for a better contribution to the AI community,” says Inoue.

These two young scholars have brought a strong sense of vigor and energy to the Institute.

"Baojian is always the last person to leave the Computer Science building each night, while Naoya has started projects with multiple of our faculty," says Professor Skiena.

Looking forward, Zhou and Inoue both say that they cannot wait to work in a more in-person, collaborative environment. The research collaborations conducted by Zhou and Inoue would have never been possible without the unwavering camaraderie and drive within the Institute. Zhou and Inoue anticipate a productive year in 2021, supplemented by collaboration and diligence.

- Alyssa Dey, Communications Assistant