DevReach 2017: Introduction to AI Using Cognitive Services

DevReach is our premier developer conference in Central and Eastern Europe. In 2017, it featured 30+ presentations delivered by 20 world-renowned speakers. These presentations were recorded and we're pleased to make them available online for everyone. The playlist for these videos is available on Youtube: DevReach Conference 2017.

Artificial intelligence (AI) and machine learning (ML) are some of the hottest buzzwords in tech. But even using a phrase like “hottest buzzwords in tech” makes us feel a little queasy - and it likely does for you, too. That kind of phrase usually suggests more hype than substance, and, often, just old wine in new bottles.

Despite their dreaded buzzword status, AI and ML can actually be useful in your day-to-day projects. And you can take advantage of what they provide without being an expert in either.

In “Introduction to AI Using Azure Cognitive Services” from Paige Bailey at DevReach 2017, you can learn exactly how to apply Azure Cognitive Services for ready-made AI.

(A quick oversimplification: artificial intelligence, or AI, generally refers to the broader category of “having computers make decisions somewhat like a human would,” or, at least, in a “smart” way. Machine learning, or ML, refers to programming computers to analyze or make decisions based on data, “learning” on its own with fewer pre-defined rules. Yes, it’s an oversimplification, but it’ll do for a start.)

We'll recap the highlights of Bailey's talk, but you can also watch a video of the presentation below:

Bailey starts with insights as to why AI is important - not just because it’s trendy but because it’s getting embedded into new frameworks. Yet there are many frameworks - and it’s hard to figure out which is best for what use, how to find the right engineers and the right infrastructure.

Thus, she explains, Microsoft as added AI as part of its Azure offerings. While you can choose the DIY model, you can also circumvent all of that by adopting Azure’s model instead.

Bailey dives deep into the DIY - or do-it-yourself - model with a detailed case study. This deep dive isn’t just a marketing case study, though, but dives into the business case and what the actual ML data looks like. This gives you a better idea of what a full, from-the-ground-up project might look like. Of course, there are a lot of time and resources involved in this option.

She then contrasts it to using someone else’s model. If you have an idea, and no allegiance to a particular model (nor the personal or institutional expertise to build your own), it may be faster to give this option a try. Plus, it’s faster to test the validity of your idea this way than to take the time to build your own infrastructure and model.

And, like so much in development now, you don’t have to reinvent the wheel. Many of the basic ML/AI capabilities have been addressed to a useful-enough level.

Bailey then gives specific examples of using three Microsoft APIs. The Computer Vision API helps give you information and context about images for better processing and curation. The Emotion API (in preview), recognizes emotions in images. The Translator Text API provides real-time language translation via a REST API call.

The overhead to try it is low: give Bailey’s presentation a look and try your ideas. You can watch the whole presentation from DevReach 2017 above or right here.

About the Presenter

Thanks to Paige Bailey for this excellent introduction to ML and AI. She’s a Senior Cloud Developer Advocate at Microsoft, specializing in machine learning and artificial intelligence.

Resources and Further Reading

• Paige Bailey
• @DynamicWebPaige (Twitter)
• Paige Bailey on GitHub