19 May Balancing the Tension Between Informed Analytics and Blackbox AI
I and in particular, deep learning, is a powerful tool for uncovering useful relationships within data; but once found, can’t explain what they mean. Contrast this with humans, armed with tribal knowledge and more traditional analytics, who understand the data relationships but just can’t find as many of them. In this episode of the IoT show I speak with Drew Conway about how to find the balance between man and machine when looking for data value.
In this episode of the IoT show I speak with Drew Conway about how to find the balance between man and machine when looking for data value.
Drew is the CEO and Founder of Alluvium where he’s responsible for the company’s vision and growth. Besides running the company, he is also a data scientist, author, and speaker.
Deep learning is all the rage in popular media but like with every other tool, it’s important to understand its strengths and weaknesses. Strength: putting structure to the unstructured nature of most data collected, especially in IoT. Weakness: can’t explain what’s been found. Strength: taking a naïve approach to processing highly variate (lots of sensors) datastores to find unexpected relationships. Weakness: can’t explain what’s been found. While it’s a powerful tool, it’s a black box powerful tool so it’s difficult (impractical) to know how it came up with its results. This presents numerous issues, nonetheless ones around understanding – the reason we’re using deep learning in the first place.
Here’s What We’ll Cover in this Episode
- Differences between AI and other advanced analytics, and which to choose, when.
- The Data Science Venn Diagram.
- Three important questions to ask before applying AI or analytics.
- Transfer learning and how it’s being used in industry.
- Why general models in machine learning are often better than the science-based analytical models when interpreting real-world phenomenon.
- The utmost importance of developing an awesome training data set.
Mentioned in this Episode and Other Useful Links
Support this Podcast
If you have been enjoying this podcast, there are a few ways you can support it:
Have an opinion? Join the discussion in our LinkedIn group
Does the black box nature of deep learning pose a problem for your company?Click here if you have an opinion on this podcast or want to see the opinion of others