Series to Highlight Women in Signal Processing: Dr. Mari Ostendorf

Dr. Mari Ostendorf is an Endowed Professor of System Design Methodologies at the University of Washington in the Electrical & Computer Engineering Department, currently serving as Associate Vice Provost for Research. She is also an Adjunct Professor in Linguistics and in the Paul G. Allen School of Computer Science and Engineering.  She received her BS, MS, and Ph.D. from Stanford University then worked at BBN Laboratories, moved to Boston University, and joined the University of Washington in 1999. During her sabbaticals, she has had the opportunity to be a visiting researcher at Johns Hopkins University, ATR in Japan, the University of Karlsruhe in Germany, the University of Edinburgh as a Scottish Informatics and Computer Science Alliance Visiting Fellow, Macquarie University in as an Australian-American Fulbright Scholar, and most recently at Google Research in London.

Prof. Ostendorf has published over 280 papers on a variety of topics in speech and language processing. For her contributions in this area, she was awarded the 2018 IEEE James L. Flanagan Speech and Audio Processing Award. In 2017, she served as a faculty advisor for the student team winning the inaugural AlexaPrize competition to build a socialbot, and conversational AI is a focus of her current work. More broadly, her research explores dynamic models for understanding and generating speech and text, particularly in multi-party contexts, and it contributes to a variety of applications, from education to clinical and scientific information extraction. She is a Fellow of the IEEE, ISCA, and ACL, a member of the Washington State Academy of Sciences, and a Corresponding Fellow of the Royal Society of Edinburgh.  She has been an active IEEE Signal Processing Society volunteer for many years, particularly in the area of publications, and in 2018 was recognized with the SPS Meritorious Service Award.

We approached her with a few questions.

Q. Why did you become a professor at the University of Washington?

I started college not knowing what an engineer was but quickly settled on electrical engineering as a major because I loved all my courses in that area. I never intended to get a Ph.D., but became excited about speech research that I learned about in a graduate student seminar series. Originally, my goal was to work in an industrial research lab after I finished my Ph.D., and it was a great opportunity for me to work with leading speech researchers at BBN after I graduated.  However, after a couple of years, I wanted the freedom to pursue more long-term research questions, and I moved to academia as an Assistant Professor at Boston University (BU). At BU, I discovered a love of teaching and working with students, and I developed an appreciation for the synergies between research and teaching. After moving through the ranks at BU, I was offered a position at the University of Washington (UW), which presented an opportunity to expand my network of collaborators. UW has been an ideal place for me to grow my efforts in language processing, connect with industry colleagues, and get involved in more applications of speech and language technology. UW has also been a great place for me because of the strong community of women in STEM and women in leadership roles.

Q. How does your work affect society?

Speech technology is always been important for accessibility, but it is increasingly being used for more convenient and efficient information access in a broad range of scenarios, whether talking to a smart device or automating phone-based information services. Spoken language technology facilitates information access in two ways: by enabling human-device interaction and by making it possible to process audio and video archives similarly to text documents. As the technology has matured, spoken language processing is having an impact in more medical and education applications, both to facilitate interactions and in diagnostic settings.

As the technology has matured, a host of new issues have arisen associated with privacy and data protection -- your voice is your identity.  In addition, like artificial intelligence and machine learning more generally, spoken language processing is subject to concerns of bias and fairness. If a user population is not well represented in the training data, the technology may not work as well for those users. These important concerns are driving new research in spoken language processing that can have a very broad impact.

Q. What challenges you had to face to get where you are today?

I started college in the distant past, so I faced my share of challenges associated with being in an underrepresented group, specifically a woman in a male-dominated field. Some things are easy to recognize, like having a TA insist on redoing all my wiring himself in a power electronics lab or being discouraged from pursuing a Ph.D. by a male colleague in the industry.  At the time I became pregnant, my institution didn't have a maternity leave policy. However, the bigger problem comes from the compounding of countless little things, microaggressions that are easy to shrug off but that eat away at your confidence and give you the sense that you don't belong. As the percentage of women in the field grows and as I broaden my network of colleagues (both men and women!), such challenges are easier to overcome. It is also important to recognize the advantages that I have had. I've been fortunate that my parents supported my college ambitions since my education set me up to pursue other opportunities. I also had some very supportive professors as an undergraduate, a terrific mentor (Robert Gray) as a graduate student, and several senior colleagues who (perhaps unwittingly) have served as mentors to me.

I have also faced a lot of the challenges that come, more generally, with being an engineer and an academic. I find that most opportunities come with challenges.  For example, moving to a new institution often means starting from scratch in building up a lab. Taking on an administrative position or a society leadership position requires learning to adjust priorities in juggling new roles and responsibilities and finding more efficient ways to keep on top of research. 

Q. What advice would you give to scientists/engineers in signal processing?

First, I'd say value community. For me, the community is my network of colleagues who I collaborate with and who I turn to for feedback on major career decisions, as well as the professional societies that I affiliate with. Being actively involved in a professional community helps you stay on top of a rapidly moving field, find mentors, and develop a broader network of colleagues outside your local community that can be useful at times of transition. As you benefit from your community, don't forget to give back, mentoring others, and supporting the next generation.

Second, when faced with a challenge, look for the opportunities it presents, and consider the advantages you do have when you are feeling disadvantaged. For example, the extra committee burden that I have had (because of good intentions to make sure women are represented on committees) gave me the opportunity to get to know some terrific leaders. In being aware of your advantages, consider ways that you can help mentor others who face related challenges.

Q. Anything else that you would like to add?

Speaking of challenges and opportunities, I'd like to make sure people know that the Signal Processing Society launched its first fully gold open access journal this year, the IEEE Open Journal of Signal Processing (OJ-SP). I agreed to serve as the inaugural Editor-in-Chief because I am excited about the prospect of increasing the impact of signal processing research through open access. Starting a new journal is challenging because authors tend to stick to the established venues. However, we've already gotten some great papers, and there are more in the pipeline. Check out the latest work and consider submitting your latest innovations to OJ-SP.