Does the Field of Big Data Mean that Machines will Replace Humans?
Companies, government, and nonprofits are spending a lot of time and money on the promise that big data can solve big problems. They’re looking to people like Dr. Wei Wang to help them. Wei is a Professor of Computer Science at UCLA and faculty in our UC Institute for Prediction Technology. She’s an expert in data mining and big data. In this week’s interview, she’ll tell you what she’s working on and the recent trends in how computer science methods can address real-world problems.
Hi Wei, can you tell us your background and about some of your current work?
My research has focused on big data analytics, bridging the areas of data mining, databases, and their applications to other disciplines. The overarching goal of my research is to achieve efficient and effective knowledge discovery on large and complex databases, and to deploy in applications with significant social impacts such as medicine, social media and networks.
When people talk about computer science and big data, we hear words thrown around like ‘data mining,’ ‘machine learning,’ ‘artificial intelligence,’ and more recently, ‘deep learning.’ Do you work in all of these areas? What are the differences between them?
Machine learning focuses on inventing new computational models for capturing patterns in data, whereas data mining focuses on addressing the scalability issue, making a model computable on big data. Artificial intelligence originated from building mathematical models to mimic human reasoning process. Deep learning refers to modeling and inferencing with multi-layer graphical models.
What are some ways your research can be applied to solving real-world problems? For example, in business, or in public sectors areas like public health?
Our research has wide applications. To name a few, we have built computational models to monitor patients with chronic diseases, to identify disease related biomarkers, to screen chemical compounds for new drug design, to analyze social media for behavior modeling and event detection.
What are some of the challenges that need to be addressed in your field? And are they able to be addressed?
There are many challenges we are tackling. The data are “big” but the space of all potential patterns is much bigger. This requires us to investigate properties held by the pattern space to guide the pattern discovery process.
Are computer scientists trying to get machines to “think” like humans or be smarter than humans? Is that the goal of data mining, and do you think that is possible?
There are some tasks that humans are far better than machines and some tasks that machines perform better than humans. What we aim for is to combine their strength, rather than having machines replace humans.