Mathematics of Data Science - Data Science is Everywhere

Описание: Data science - what's under the hood? This animation, from the SIAM Journal on Mathematics of Data Science, explains that data science really is EVERYWHERE!

SIAM Journal on Mathematics of Data Science (SIMODS) publishes work that advances mathematical, statistical, and computational methods in the context of data and information sciences. We invite papers that present significant advances in this context, including applications to science, engineering, business, and medicine.

---
FULL MANUSCRIPT:
Right now, you’re a few clicks away from streaming a 4K video tour of a far-away city, and exploring a 3D map of it in virtual reality. If you want to actually visit the city, your phone can arrange for a car — maybe even a self-driving car — to pick you up just as you land. While it’s shuttling you around, apps can suggest hotels and sites to visit.

We are living in the age of data science.

Data science is everywhere, but how does it actually work? When the data analysts, scientists, and engineers who build these applications run up against the limits of what’s currently possible, how do they make the next breakthrough?

The Society for Industrial and Applied Mathematics has a new journal for mathematicians, computer scientists, geneticists, neuroscientists, economists and anyone who works with big data: the SIAM Journal on Mathematics of Data Science, known as SIMODS.

Through SIMODS, researchers are popping the the hood and tinkering with the engine that makes these applications work, and work better: applied mathematics, and the related domains of computer science, statistics, signal processing, and network science.

The compression techniques that allow you to stream a 4K movie are in a constant race with growing file sizes. In the future, techniques like matrix sketching can be used to efficiently discover the underlying low-dimensional manifold and achieve even greater compression rates. This will make your movies stream faster and with better image quality.

Deep learning techniques use stochastic optimization for quick and accurate translations. Even more powerful techniques will be necessary to handle the technical language found in specialized categories of speech, like those in law, medicine, and science.

What about unsupervised learning, where there are no categories at all?

Would you trust your computer to organize the photos from your trip, with no instructions on what folders to make? What about images of brain scans, and your computer could find never-before-seen patterns and correlations that human neuroscientists would never think to look for? Applied math techniques like clustering can make these organizational tasks even better, allowing for applications that seem like science fiction today.

Looking forward, imagine machine learning methods that can keep your data completely private, explain their decisions while offering customized suggestions, and be robust to new situations. Can data science move us forward in terms of fairness and diversity? What about using algorithms to achieve long-term goals?

Computer scientists and engineers are inventing the future every day, and applied mathematics gives them the tools they need to keep moving forward. SIMODS is looking for interdisciplinary work that pushes the boundaries of data science and takes the field in new directions.