Applied AI News 3

Main message

When it comes to applications of AI and Machine Learning these days, we are mostly limited by our imagination, which at its core is limited by our exposure to possibilities. The intention of this series of posts is to make the audience more aware of how AI and Machine Learning are leveraged by industries, exposing the reader to new applications, and extending the opportunities one could imagine in their daily business.

Summary of the week

AI model simplification, encoding human relationship, COVID19 and AI, democratizing AI, AI-specialized hardware, psychology, and programming are the headlines of this week.

• Deep networks simplification: It is well-known that deep neural networks involve a large number of connections, making it challenging to deploy on hardware, and make them complex (hence, more likely to overfit). New ways for simplifying these are available.
• AI in the human relationship: Conversations carry a big deal of information about the relationship between the people. An AI has been trained to abstract the relationship dynamic between people using their conversations.
• AI and disease: COVID-19 has impacted many people, but one main population in the focus is the elderly. Given all the social distancing and isolation, the elderly can feel lonely and may need a more dedicated monitor. AI has been designed to accomplish this task.  
• Democratizing AI: Democratizing AI? Here is a new viewpoint based on actual democracy that I personally like (leave comments if you think otherwise).

Video of the week: Elon Musk and Jack Ma, interesting AI debate. I personally like Ma better. What do you think? Discuss in the comments section.

• AI Hardware: Distinction between memory and compute in computers is a long-standing design choice, either in personal computers or in Cloud architectures. But is it efficient? Of course not as it requires migrating data from one memory to the other to make computations more effective. This has been challenged recently by IBM, an attempt to build a device that uses the same hardware for computing and memory.
• AI and psychology: Personality from selfies, I am not sure if I can do it. It seems like there is an AI that can beat humans in that task. 
• Movement: Detecting hand and finger movement has applications in seizure detection, mood detection, and remote human-computer interface. A new AI-based method can detect this accurately using a new sensor.
• AI and programming: 70 bugs per 1000 lines of code is an alarming but real statistic, each bug taking 30 times more time to fix than writing the original code, emphasizing how difficult is this task. It seems like AI can help here.

-------------------------------------------------------------

That is it for this week. There is a lot happening in the area of AI and Machine Learning, we just need to find the opportunity, the tool is there. Like always, comments are very welcome. 

Best wishes and have a great week/end.

Reza 

rezabny@gmail.com