What is Your Interest in AI and Machine Learning?

anyone can learn basics of AI and ML.

coursera or other similar sites are great

interesting

In AI do the functions and algorithms change based on the data?

I wanna know too... gonna ask a friend

I am doing AI at an application level here and there, but most of the time still PC based, although trying hard myself doing it on embedded platform. So far the only embedded platform that i have done is on NVIDIA Jetson TX2 with a yolo net object reconigtion, but at the moment is attempting to do it on FPGA.

I’d definitely read or watch anything published in this area. I’d been tempted to do something for the connected micro controllers challenge (and maybe still will), but time and commitments to existing projects mean it never quite gets to the top of the list. Definitely a popular topic at the moment!

Applying Machine Learning on microcontrollers / Raspberry Pi will bring more capabilities for IoT and enable Edge Computing.

AI with edge devices like Raspberry Pi and other low power devices will definitely help makers. Even some of the microcontrollers are coming with hardware accelerated neural nodes, that would help in making smart devices.

There are already some tools like TensorFlow available to bring AI to devices like Raspberry Pi.

AI has been around for more than 40 years, going through many phases of hopes and disillusions. It is rather a vast area, but the new boost of successful research and interest in the academic and industrial sectors is mostly due to the progresses in the branches of Machine and Deep Learning techniques. The picture below shows how AI, Machine Learning and Deep Learning relate to each other (in a simplistic form):

There are several factors that have contributed, and are essential, to the progress made in the this field:

- the tremendous improvement in processing power (thanks to the Moore's Law)

- the availability of a vast amount of data to be consumed (the famous Big Data)

- the improvement of the techniques used to process the data

Personally, I believe the most fascinating and exciting area, and the most active currently, is the study and application of Deep Learning (DL) techniques.

Unlike the typical AI approach of using some "pre-coded" logic, to solve problems, Deep Learning techniques leverage the ability of some neural network to "learn" how to solve a problem by using training datasets.

It is definitely a fascinating field, and it can offer a plethora of possible applications to solve many different problems in many diverse fields.

To anybody who likes to start learning about all this, Andrew Ng's courses on Coursera (Machine Learning base courses and the Deep Learning specialization courses) are great, although they do require some commitment (time and money).

Alternatively, there are several books published by Packt Publishing on the topic, which are available for free (although they require you to sign up to their website). You can find some of the books here (check in the "Free Machine Learning eBooks" section).    

And, if you fancy to get your hands dirty, there are some DL frameworks you can use to implement your first DL project. The one I found easier to start with is Caffe | Deep Learning Framework. It can be used with Python, which makes a good rapid development environment.

Fabio

P.S: I have a vested interest in this area, as currently I'm doing some R&D on Deep Learning applied to prediction algoritms...

I recently attended a presentation where the speaker started off with definitions for AI and machine learning. They are different by her definition.  She focused on the consequences of failures and how to address them rather than the percentage success.  Very interesting.

I have followed developments with interest but have only superficially used AI outside of consumer applications like voice assistants.  The easier to use approaches seem limited and the more powerful approaches seem - well I am lazy and have limited resources and time.  I would follow anything published with interest especially if it had an interface to the physical world and microcontrollers.

Something I've never read too much into but would really like to learn some more. The bit I really need to understand is how AI software differs from, what many of us are familiar with, the procedural programming paradigm, and what makes it AI rather than just a really complex and convoluted collection of algorithms. In AI do the functions and algorithms change based on the data? I keep thinking back to what makes intelligence andeven  more, the human being....our ability to learn from what we do, the actions and reactions and the ability to just do things out of the ordinary. [At that point I considered writing something odd to prove my point, thought about it, considered the feelings of other E14 members, and then decided to write this sentence instead ! ]

Rod