The many new and sometimes abstract AI terms can sometimes be a little intimidating for beginners. Even when it comes to explaining the difference between machine learning and deep learning, confusion can sometimes arise. To simplify things, we’ve come up with a conversation between two fictional characters, student Emma and expert Dr. Tom Neuron.
Questions & Answers: Machine Learning vs. Deep Learning
Disclaimer: The following talk never took place and the people taking part and their names are fictitious. It was created to make it easier to immerse yourself in the topic.
Stacked Venn Diagram for: AI – Machine Learning – Deep Learning
-
Emma:
One question has been on my mind for some time now, that is: How is machine learning different from deep learning?
-
Dr. Tom Neuron:
Thank you Emma, that’s a very interesting question. To get a basic understanding of how AI, machine learning and deep learning relate to each other, I have the following illustrative comparison for you. Imagine you are looking at a frying pan from above when you have finished frying an egg.
Then you can see the entire surface of the frying pan – that’s the AI. Now there is the fried egg with the egg white and the yellow egg yolk.
You can equate the egg white with machine learning.
And deep learning is the egg yolk. Deep learning is always machine learning. But machine learning doesn’t always have to be deep learning. Both, Machine Learning and Deep Learning, are part of AI. -
Emma:
I’m not so good at remembering things like that. Do you have a graphic with the subsets?
-
Dr. Tom Neuron:
Please take a look at the following Venn diagram as a fried egg on the inside of a frying pan. You can use the diagram to visualize the subsets concisely: The egg yolk (deep learning) is part of the egg white (machine learning), and the fried egg (machine learning including deep learning) is part of the inner surface of the frying pan (artificial intelligence). The image of the fried egg in the frying pan is very memorable 😉
Machine Learning vs. Deep Learning: Comparison Table
-
Emma:
Ah, great! Now I finally understand all these AI topics better. This is followed by another question: What are the concrete differences between machine learning and deep learning? I understand that deep learning is a component of machine learning. But how do machine learning and deep learning differ from each other? There must be differences, right?
-
Dr. Tom Neuron:
This is a very important question There is not just one difference, but several distinguishing features between ML and DL.
To delve deeper into the topic, you need to know these differences between machine learning and deep learning. I have created a comparison table for important distinguishing features.
To make you a little curious about the differences: Deep learning algorithms are so complex that they are often difficult to understand and explain. And here is the table.
| Criterion | Machine Learning | Deep Learning |
|---|---|---|
| Data processing | Requires structured data | Can process unstructured data (images, speech, text) |
| Data requirements | Less data required per feature | Requires very large amounts of data per feature |
| Model complexity | Simpler models (e.g. decision trees, SVM) | Very complex models with many parameters |
| Structured data required? | Tabular data, structured data, simple pattern recognition | No structured data required. E.g. image, speech and pattern recognition at a high level |
| Computing power | Low computing power required (CPU) | Very high computing power required (GPU/TPU) |
| Neural network with how many layers? | 2 to 3 layers | ≥ 4 layers (incl. input and output layer) |
| Is human intervention necessary for model training? | Yes: Model training requires intervention during feature extraction, specification of weightings | No: Runs automatically, therefore no intervention necessary, model learns itself via feedback from known errors |
| Comprehensibility | Models are often explainable and comprehensible | Hard to understand “black box” models |
Sources:
-
Emma:
Do you have any further sources like online expert articles to delve much deeper into the comparison of machine learning and deep learning?
-
Dr. Tom Neuron:
Yes, I have these sources for you from AI expert websites that I have compiled over time.
https://cloud.google.com/discover/deep-learning-vs-machine-learning
https://aws.amazon.com/de/compare/the-difference-between-machine-learning-and-deep-learning
https://www.ibm.com/think/topics/ai-vs-machine-learning-vs-deep-learning-vs-neural-networks
https://www.salesforce.com/artificial-intelligence/deep-learning-vs-machine-learning