March is the National Women’s History month. What better way to celebrate than attending the Women in Data Science Conference (WiDS)? MBS students remotely attended this unique and energetic event along with members of more than 150 organizations across the globe. The fourth annual Women in Data Science (WiDS) conference was hosted at Stanford University on March 4th, 2019. Talk topics ranged from ethics, cybersecurity, data privacy, ML applications to career motivation. WiDS conference speakers came from a broad spectrum of backgrounds; each speaker had found data science either as a student or on stumbling upon it by chance in mid-career. We are inspired by and have learnt from these women data scientists who have broken barriers, overcome obstacles and paved the way for future female Data Scientists.
The conference keynote speaker, Padmasree Warrior (CEO NIO), emphasized the role of data science in progressing humanity. She outlined how we have moved on from desktop or device based towards wearable technologies. Her predictions for the future considered blockchain, autonomous vehicles, healthtech and the emergence of new computing technologies. With such a huge wave of technological revolution awaiting, understanding the value of data is key to the governments from around the world developing new policies to accommodate the future events. In her closing remarks she said, “Keep your eyes on the stars and feet on the ground,” referring to having a grand vision but a detailed execution plan.
One of the most design-oriented speakers at the WiDS conference was Hilary Parker (Stichfix). Though a molecular biologist by training, she has embraced the value in design thinking. She recommends the book ‘Design Your Life’ by B. Burnett and D. Evans about how to build a well-lived joyful life. Through her professional experiences, she has come to believe that everything is a system and, in designing these systems, most time should be spent in data collection followed by data munging then modeling. Product design is no mystery, rather it is a discipline according to Parker. The best product design can be attained through learning by doing and cultivating empathy.
Healthcare Data scientists Marzyeh Ghassemi (University of Toronto) and Yoky Matsuoka (Nest) are very passionate about what it means to have healthy healthcare. The healthcare data collection process is expensive, tricky and biased. Furthermore, Marzyeh said that Data + ML ≠ Insights. Algorithms, as in machine learning (ML), can give wrong answers. To achieve optimal results, we have to eradicate biases and find ways to use self-reported and passive data along with expert health records. Yoky thinks that the tripod solution of tech companies, healthcare providers and insurance companies can usher in the new era in healthcare. Both women emphasized that early engagement is essential for healthy long-term outcomes.
Emma Brunskill (Stanford University), Anima Annankumar (CalTech) and Madeline Udell (Cornell University) emphasized the importance of using Reinforced Learning, structure in data and generalized low rank models. These talks highlighted how data scientists use creative yet robust methodologies to solve problems of human-in-the-loop systems, safer autonomous systems and filling in missing data. They also recommended searching out ready-to-use tools available in python and R rather than developing algorithms from scratch.
Timnit Gebru (Google) and Cynthia Dwork (Harvard University) spoke about limitations of AI and the use of differential privacy guidelines for preserving data privacy. They understand that privacy and generalization are aligned, hence measures have to be taken to protect the outliers. For example, police actions, population analysis for policy decisions or funds allocation and hiring processes have flawed social, cultural and language biases which cause ethical issues. Data sheets and model sheets as well as resilient and graceful techniques can help solve some of these issues.
Alicia Carriquiry (Iowa State University) and Laura Kegelmeyer (Lawrence Livermore National Lab) gave very interesting talks about pattern matching in crime forensic evidence analysis and Optics damage/repair inspections. Human analysts approximate possible matches in the patterns to find connections, however the ML algorithms look for exact matches. Both women used Random Forest to automate the analyses. The automation showed that Similarity ≠ Source.
The career panel conversations were relaxed and honest. Yinglian Xie (Data Visor) talked about her mission to detect cyberfraud. She found success in her start-up experience by following her passion, having a solid business plan and finding a genuine partner. Natalie Harris (BrightHive) and Emily Sands (Coursera) with backgrounds in sociology and labor economics shared their passion of using data science to bring social impact solutions. Panelists excitedly talked about matching your unique strength to problems you want to solve, ask questions about the systems you want to resolve, celebrate failure, cultivate awareness and be empathetic! Hearing about the panelists’ Data Science journey and the many paths to true success is encouraging.
The theme of cultivating foundational knowledge along with soft skills, which is essential to progress in the field of Data Science, was very evident in each of the talks and conversations. Rutgers’ PSM students experience these in their required and elective courses on a daily basis. ‘What you see, you can be!’ was truly exemplified at this conference. Above all, the presence of many female role models in Rutgers’ PSM program as at WiDS are continually making students’ Data Science journey wholesome!