in

Annual Women in Data Science conference discusses fake news

What do radiation waves from space, the U.S. Census, and the human genome have in common? All three, like so many things today, involve massive amounts of data. These data can unlock insights and lead to new solutions and better decision-making — for those who have the knowledge and tools to analyze it.

The impressive variety of applications for data science tools and techniques were on display at the Women in Data Science Conference (WiDS Cambridge), held at the Microsoft NERD Center in early March, before MIT and the Commonwealth of Massachusetts began to de-densify in response to the Covid-19 emergency. Co-hosted by the Institute for Data, Systems, and Society (IDSS), the Harvard Institute for Applied Computational Science, and Microsoft Research New England, WiDS Cambridge is one of dozens of satellite WiDS events around the world. The program showcases women who are not only using data science tools in their research or business, but who are leaders refining those tools and recruiting more women into the field.

The day’s signature event was a panel discussion on data science and fake news called “Data weaponized, data scrutinized: a war on information.” The panel was moderated by Manon Revel, a doctoral student in the IDSS Social and Engineering Systems (SES) program whose research has analyzed popup ads to see how exposure influences readers’ assessment of news credibility. Addressing current challenges, Manon shared: “Understanding the effect of false information and combatting it is crucial. It requires thinking through the technology design, but also the regulatory framework and the political and social context.”

The panel also included Camille Francois, chief information officer for Graphika, a social network analysis startup that uses AI to understand online communities. “We don’t know how to measure the impact of foreign interference for many complicated reasons,” said Francois. “The aim of a foreign interference campaign is not necessarily to impact a vote. It’s to divide, it’s to confuse, and it’s to create chaos. How do you measure chaos?”

In addition to the discussion on misinformation, WiDS Cambridge featured a wide variety of insights from industry and academia. Asu Ozdaglar, deputy dean of academics, head of the Department of Electrical Engineering and Computer Science, and faculty member in IDSS and the Laboratory for Information and Decision Systems (LIDS), highlighted robustness in machine learning. Citing the common example of image classification system errors, she explored how ‘perturbed’ data can, with small variations, disrupt otherwise accurate models, and offered a ‘minmax’ approach using generative adversarial networks (GANs) to increase robustness.

For an industry perspective, Jess Stauth, managing director at Fidelity Labs, provided ways to apply basic research principles to modern data science business problems. Data science is a collection of tools from statistics to computing, she says, and businesses require infrastructure to use them to create tangible business value. “A data scientist alone in a room with a laptop is probably not going to be all that successful,” she muses.

The conference provided opportunities for participants to network and job search, with sponsor companies hosting recruiting tables and answering questions. WiDS also empowered newer practitioners with a student and postdoc poster session and lightning talks. Over 30 poster presenters participated, showcasing work in fields as diverse as demographic bias in natural language processing, crime prediction, neurodegenerative disease, and sustainable buildings.

“WiDS is a wonderful event where you can interact with your peers, present your research, and build confidence,” says Marie Charpignon, a graduate student in MIT’s SES PhD program who presented a poster on using causal inference on electronic health records to explore repurposing diabetes medication to treat dementia. “The conference brings together students, professors, industry researchers, and even venture capitalists in search of promising ideas. WiDS gives you a sense of the myriad paths you could take after graduation.”


Topics: Institute for Data, Systems, and Society, Electrical engineering and computer science (EECS), Laboratory for Information and Decision Systems (LIDS), Special events and guest speakers, Data, Analytics, Women in STEM, MIT Schwarzman College of Computing, Machine learning, Technology and society, Social media


Source: Data Management & Statistic - news.mit.edu

DHS CISA: Companies are getting hacked even after patching Pulse Secure VPNs

Coronavirus scams: This is how much people have lost to online fraudsters so far