Dr Shirin Glander Setup All analyses are done in R using RStudio. For detailed session information including R version, operating system and package versions, see the sessionInfo() output at the end of this document. All figures are produced with ggplot2.Continue reading… Building meaningful machine learning models for disease prediction
Increasingly, connected systems and devices are generating data that produce insights for improving business processes and consumer experiences. IDC predicts that by the year 2020 there will be 44 zettabytes (that’s 44 x 10) of information, spawned partly by consumerContinue reading… Leveraging Data Analytics and Internet of Things in Your Digital Transformation
By Jane Wakefield Google’s artificial intelligence unit DeepMind is getting serious about healthcare – with ambitious plans to digitise the NHS – but first it needs to convince patients to hand over their medical records. Back in February, it beganContinue reading… Google Deepmind: Should patients trust the company with their data?
Today, businesses can collect data along every point of the customer journey. This information might include mobile app usage, digital clicks, interactions on social media and more, all contributing to a data fingerprint that is completely unique to its owner.Continue reading… 5 Benefits of Data and Analytics for Positive Business Outcomes
What are Tree Methods? Tree methods are commonly used in data science to understand patterns within data and to build predictive models. The term Tree Methods covers a variety of techniques with different levels of complexity but my aim isContinue reading… Making data science accessible – Machine Learning – Tree Methods
IN BRIEF Machine learning system DeepCoder can locate and piece together code from existing software to write new programs up to around five lines of code in length. Eventually, DeepCoder could become a valuable tool for programmers by tackling theContinue reading… Our Computers Are Learning How to Code Themselves
Imagine a world where a sick child can “visit” a pediatrician from the comfort of their own bed, while their parent has the prescription filled without ever setting foot in a pharmacy. Such convenience will soon become a reality forContinue reading… How IoT is Creating Investment Opportunity in Healthcare
Potentially describing how general artificial intelligence will look like. Since scientists started building and training neural networks, Transfer Learning has been the main bottleneck. Transfer Learning is the ability of an AI to learn from different tasks and apply itsContinue reading… DeepMind just published a mind blowing paper: PathNet
It’s not hyperbole to say that use cases for machine learning and deep learning are only limited by our imaginations. About one year ago, a former embedded systems designer from the Japanese automobile industry named Makoto Koike started helping outContinue reading… How a Japanese cucumber farmer is using deep learning and TensorFlow
Math. Math. Oh and perhaps some more math.
That’s the gist of the advice to students interested in AI from Facebook’s Yann LeCun and Joaquin Quiñonero Candela who run the company’s Artificial Intelligence Lab and Applied Machine Learning group respectively.
Tech companies often advocate STEM (science, technology, engineering and math), but today’s tips are particularly pointed. The pair specifically note that students should
eat their vegetables take Calc I, Calc II, Calc III, Linear Algebra, Probability and Statistics as early as possible.
From this list, probability and statistics are perhaps the most interesting. From what I remember about high-school, those two subjects are regularly dismissed as too-obvious strategies for skirting the informal AP Calculus preference of top colleges and universities (AP Statistics is often thought of as a cop-out by students).
If differential equations represents the electricity that powers machine learning, statistics represents the gears of the machine itself — as the company touches on in a series of AI explainer videos we linked to at the bottom of this post.
To be fair, LeCun and Candela are most likely addressing the college crowd, though its important to consider incentives across all levels of education. Simply, we all could probably use some more statistics in our lives. Beyond math, the two say
more math engineering, computer science, economics and neuroscience are also important subjects in today’s economy. How else would a fledgling machine learning student learn to leverage neuroeconomics and cognitive bias to target ads?
The pair also point to philosophy as a necessary prerequisite to understanding knowledge and learning. Amidst all the talk of News Feed bias, it’s important to remember that there is a human behind every application of machine learning. We don’t yet know how to escape the black box problem, but we do know that it will be humans working to figure it out and it would sure help if those humans understood how learning works before they start manipulating data.
Lastly, Facebook turns its attention to the actual mechanics of getting a job in the field of machine learning. Most of these tips are self-explanatory: find a professor to work with, consider working with PhD students who have more time on their hands and try to secure an industry-focused internship regardless of your future aspirations to understand how AI works in the real world.
When applying to PhD programs the two note that being able to identify a professor you want to work with is far more important than program ranking. Once there, students should work to address a specific problem and try to release a piece of open source code before all is said and done.
For more, click here.