The Difference Between A.I. and Machine Learning and Deep Learning

30 thoughts on “The Difference Between A.I. and Machine Learning and Deep Learning”

1. If you enjoyed this you may want to find out just how powerful A.I. can be with this video >>> https://www.youtube.com/watch?v=_JBQYVXkTfg&t=2s

2. Thanks for breaking this down!

3. That's great, thanks for breaking it down! What source do you suggest to get more information on this topic?

4. Deep Learning < Machine Learning < Artificial Intelligence

5. This is a nice overview indeed. Having stated that ML is a subset of AI, it could be interesting to indicate what else is there in AI, i.e. what is the subset AIML. This is where many beginners have problem. Similarly, what is there in MLDL?

6. I was hoping for a clearer explanation of the difference between AI and Machine Learning. Could you describe what it is that makes them different? I have always seen AI as a subset of ML, not the other way round… ie ML is anything that uses an algorithm to train on a data set in order to perform a task and AI … based on the perceptron … specifically achieves this task with a neural net. What do you think?

7. Concise and very effective in this short time: brilliant!

8. When machines start to do what humans can do, then we say that they have artificial intelligence. The method through which this is achieved is majorly through machine learning. Even in machine learning algorithms, a set of deep learning algorithms are performing exceedingly well which is the cause of the current rage of AI field

9. Great. This topic needs a lot more clarity in most discussions by generalist managers, especially regarding what is realistically achievable and what not. There is a lot of hype around AI, ML and Big Data. Especially vendors often overstate the benefits.

10. That was brief and helpful.  Thank you.

11. Well explained

12. Excellent video. Really simplifies things. In a world full of complex theorical definitions, this video shows the light.

13. Thanks, for the initiative. There is a lot of hype in the market and terms are often misused. It gives clarity. Sharing my point of view
    1. Deep Learning is not the subset of Machine Learning. Deep learning and Machine Learning are two different approaches.
    2. Deep Learning uses network architecture and does not need feature engineering.
    3. Machine learning, it is more using statistical model and feature engineering necessary to get better results.
    4. AI is the mother of all type of learning that simulate “human intelligence” in terms knowledge, reasoning, planning, problem-solving, learning and perception

    Recent days, Cognitive learning is being used. I understand it is the application interface of AI, to have interaction with the human in a natural way – Speaking, Listening, Writing, Reading, See and Project. Behind Cognitive learning, it may use AI/Machine learning and Deep learning methods.

    Open to suggestions to learn more on this subject.

14. Very useful

15. Thanks for making the difference simple to understand

16. There are a lot of courses online offering AI, machine learning and deep learning seperately.. I am curious about deep learning… Could I start learning directly from deep learning without any knowledge of AI and machine learning?? or should I have to learn the AI and machine learning and then deep learning because it will take a lot of time and money

17. AI ⊃ ML ⊃ DNN

18. Hi Bernardo, are you aware of Brainchip Holdings https://www.brainchipinc.com/ , and in particular their patented Akida technology that learns autonomously, evolves and associates information just like the human brain, soon to released as a silicon chip. The inventor Peter Van Der Made spoke at the recent AGM and revealed (in no particular order) The AkidaTM device is a complete Neuromorphic System‐on‐Chip (NSoC)
    * There will not just be one Akida Chip.*** Newer Akida Chips will be designed and released at later dates for special applications and devices.
    Image recognition Accuracy: 90% vs CNNs 80% humans 85%
    The big difference between Akida and Deep Learning is speed. Speed is so important. It's a matter of Life and Death. In an autonomous vehicle if you are relying on Deep Learning you will be lunch. BA processes 600 frames per second. That's is only a fraction of the time that deep learning takes.” "Deep learning has no concept of time. Akida does."
    Low power consumption vs GPUs.
    The Akida Chip is being designed to incorporate the large dataset libraries used in training deep learning systems. Also he advised that the SNN blocks are designed to use all deep learning developments.
    Autonomous vehicles employing Akida IP there will be no need for connection to the GPU, no Nvidia, no cloud and no latency. In an Akida Autonomous Vehicle environment there will be a central nervous system of say 16 chips, and each peripheral sensor will be connected to a chip.
    Deep Learning is dead and he does not see deep learning having any place at all in Autonomous Vehicles. Something to do with "back propagation" See HotCopper post 33046468 by Cyberworkshttps://hotcopper.com.au/threads/ann-agm-presentation.4183123/page-64?post_id=33046468
    Akida does listen to sound. For instance it is quite capable of monitoring the health of an aircraft engine while operational.
    There is an artificial retina in Brainchip Accelerator and Akida.
    No programming, less data required, less processing power, less storage and less costly to implement than other machine learning models including those implemented on CPUs or GPUs.
    Porting existing Convolutional Neural Networks (CNNs) to SNNs Simulated Neural Network opens a large and immediate market while other learning modes (Autonomous Supervised and Unsupervised) support future use cases.

19. Go in another place to sell your course!

20. AI superset of intelligent activities, ml subset and DL most subset (most complex tasks)

21. I will also add a classification: Symbolic AI and Sub-symbolic AI. Symbolic deals with explicit representation of knowledge whole sub-symbolic (which is otherwise called Artificial neural networks) represents knowledge as differences in weight values between nodes in different layers. The final layers contain classification / identification nodes for the possible outputs of a task. Machine learning is one of the techniques of Artificial Intelligence where the application is able to better its knowledge and thereby performance. Again there is symbolic ML and sub-symbolic ML. Symbolic ML refers to learning concepts, rules, clustering objects, finding analogies using techniques of induction, abduction and analogy. All sub-symbolic systems are inherently learning; as the weights between nodes are learnt to finally arrive at the correct output. One form of sub-symbolic (ANN) is Deep learning. Initially ANNs had only a few layers of nodes. Increasing the layers of nodes lead to a problem of vanishing gradient (the error is being minimized gradually using a gradient descent: sliding down an error sloe until we reach a zero error. Unfortunately as the number of layers of nodes increases this slope changed too little; almost stayed flat. This problem was solved by Geoff Hinton which has resulted in ANNs with multiple layers of nodes called Deep Learning.

22. Loved the explanation.

23. 2:51 better THAN us

24. great video info was on point

25. Good video, wished it had more details

26. Whatd ANN and how its releted to machine and deep learning

27. Download our free list of the worlds best AI newsletters ? https://hubs.ly/H0dL3qz0

28. * in Short: AI>ML>DL
    * in Detail: AI is of a broader scope it has many sub areas within it and ML is one out of those many areas, Why people don't talk about other areas is because ML has had the most impact on our lives as of now than any other sub area out there. Now coming to Deep learning, Its a subset of ML, and again why Deep learning is heard more is because its this subset has had the best impact as of now when compared to other areas of interest.

29. great explanation, thanks

30. you've earned my subscription in 3 mins. Best concise and visual explanation of anything ever.

Comments are closed.