For this year's Episerver Ascend conference, I gave a talk on machine learning. I wanted to share some info about the talk, as well as the slides I used (Slides at the bottom).
I won't go into too much detail about the talk, but I wanted to summarize some of the topics.
What is machine learning
I went through some info around machine learning. Like what it is/isn't. Some common terms, as well as the different categories. I also listed some examples of how machine learning is being used today. You can find references to those examples below.
Why Should I care?
I went into some reasons why machine learning is relevant. Even to those who already use personalization. The main takeaway being that machine learning allows for personalization at scale.
Machine Learning & Episerver
I then went into what Episerver is doing with machine learning.
I touched on things like Personalized Find for personalized search results. Personalized recommendations, both product and content, with Episerver Perform and Profile Store. Personalized email campaigns using Episerver Reach. And using Insights to visualize user data and create custom segments.
I then went into tools that will allow you to expand outside of what Episerver provides by default. I focused on Microsoft tools because we are already in their ecosystem. I covered more development related items such as ML.Net and Azure Cognitive Services. I also went into items for business users, like Azure Machine Learning Workbench.
On slide 9 of the deck, I referenced Netflix's use of machine learning. Among other things, I called out the use of personalization when creating thumbnails. You can checkout Netflix's machine learning blog, as well as the personalization post below.
On the next slide, I referenced AirBnB and how they use image classification. They also have a cool tech blog on medium, and I added the specific article for image classification as well.
The last item I wanted to reference was a story about how Target was able to target a customer as being pregnant. It was unique in the fact that she did not inform anyone of the pregnancy yet. Since she was under age, her father was angry at the local Target manager after she received coupons for things like cribs, bottles, etc.
Its an interesting story about how retailers like Target are leveraging machine learning. You can read the full story in this Forbes Article.
I went through some tools that Microsoft provides to get you up to speed with Machine Learning.
After the talk some people had some follow up questions.
What do I do if I don't have enough data to get started with Machine Learning?
I referenced the need for data a lot in my talk. On slide 23 I brought up this exact question. It's considered by organizations as a barrier to entry for machine learning.
I advised starting now by working with Google analytics and Episerver Insight. User data is necessary for things like personalization. I called out some tools that Microsoft Cognitive services provides. Things like image recognition, sentiment analysis, and several others don't always need user data.
There are a lot of Microsoft tools you mentioned, which should I start with?
That's a tough question to answer. I mentioned in my talk that its important to have the right question in mind first. You shouldn't try to shoe horn machine learning into your solution for the sake of it. So you can use tools like the Microsoft decision tree I liked above to help with that.
If you want to get into machine learning for fun, I'd recommend Microsoft Cognitive Services. They do a lot of the heavy lifting and you only need to provide data. So its a good place to start.
This was my first talk at Ascend. I appreciate Episerver giving me the chance to speak about such an interesting topic.
If you have any questions related to my talk, or follow up questions to the topics covered, feel free to contact me