The best AI/mL roadmap for beginners

Piton

Python is the gold standard and programming language for automatic learning and ai.

Beginners are often trapped in the so -called “best way” to learn Python. Any introductory course will be enough, since they teach the same things.

The main things you want to learn are:

• Native data structures (dictionaries, lists, sets and tuples)
• For and while loops
• IF-ELSE conditional statements
• Functions and classes

You also want to learn specific scientific computer libraries such as:

• NUMPY– Numerical computing and matrices.
• Pandas– Data manipulation and analysis.
• Plotlib food AND TRUST– Data visualization.
• Lear  – Classic ML algorithms implementation.

Resources:

TIME LINE:Again, depending on their background, this should take a couple of months. If you already know Python, it will be much faster.

Data and algorithms structures

This may seem a bit out of place, but if you want to be an automatic or ai learning engineer, you must know structures and data algorithms.

This is not just for interviews; It is also used in ai/ml algorithms. You will find things like recoil, depth search and binary trees more than you think.

The things to learn are:

• Linked matrices and lists
• Trees and graphics
• Hashmaps, tails and batteries
• Classification and Search Algorithms
• Dynamic programming

Resources:

• Neetcode.io– Great introductory, intermediate and advanced data structure and algorithms courses.
• LEET Code AND Trick– Platforms to practice.

TIME LINE: About a month to nail the basics.

Automatic learning

This is where fun begins!

The four previous steps involved preparing their basis to address automatic learning.

In general, automatic learning is in two categories:

• Supervised learning– Where we have target labels to train the model.
• Not supervised learning – When there are no target labels.

The following diagram illustrates this division and some algorithms in each category.

The algorithms and key concepts that must learn are:

• Linear, logistics and polynomial regression.
• Decision trees, random forests and gradient trees.
• Support vectors machines.
• K-Media and K-get neighboring clustering.
• Features engineering.
• Evaluation metrics.
• Regularization, bias against the compensation of variance and cross validation.

Resources:

TIME LINE:This section is quite dense, so it will probably take approximately ~ 3 months knowing most of this information. Actually, he will have really dominated everything in those resources.

ai and deep learning

There has been a lot of advertising around ai since Chatgpt was launched in 2022.

However, ai itself has existed as a concept for a long time, which dates back to the 1950s, when the The neuronal network originated.

The ai ​​to which we refer at this time is specifically called the generative ai (Genai), which is actually a fairly small subset of the entire ai ecosystem as shown below.

As the name implies, Genai is an algorithm that generates text, images, audio and even code.

Until recently, the landscape of ai was dominated by two main models:

However, in 2017, an article called “Attention is all you need” It was published, presenting the architecture and Transformer model, which since then has replaced CNNS and RNNS.

Today, the transformers are the backbone of large language models (LLM) and unequivocally govern the landscape of ai.

With all this in mind, the things you should know are:

• Neural networks–The algorithm that really puts ai/ml on the map.
• Convolutional and recurring neuronal networks –It is still used quite today for your specific tasks.
• Transformers –The current state of art.
• RAG, vector databases, llm tuning fine –These technologies and concepts are crucial for current ai infrastructure.
• Reinforcement learning– The third type of learning used to create as well as ALFAGO.

Resources:

• Deep learning specialization byAndrew Ng. -This is the monitoring course of automatic learning specialization and will teach everything you need to know about deep learning, CNN and RNN.
• Introduction to LLMS By Andrej Karpathy (former senior director of ai in Tesla) –Get more information about LLM and how they are trained.
• Neural Networks: Zero A Hero–Start relatively slow, building a neuronal network from scratch. However, in the last video, it makes you build your own previously trained generative transformers (GPT)!
• Reinforcement Learning Course – Conferences from David Silver, a principal researcher in Deepmind.

TIME LINE:There are many things here and it is a rather hard and avant -garde call. So, about 3 months is probably what will take you.

Mlops

A model in a Jupyter notebook has no value, as I said many times.

For your ai/mL models to be useful, you must learn to implement them in production.

The areas to learn are:

• Cloud technologies such as AWS, GCP or Azure.
• Docker and Kubernetes.
• How to write production code.
• GIT, Circlei, Bash/ZSH.

Resources:

• Practical mlops (Affiliate Link)-This is probably the only book you need to understand how to implement your automatic learning model. I use it more as a reference text, but teaches almost everything you need to know.
• Design of automatic learning systems (Affiliate Link)– Another great book and resource to vary your source of information.

Research work

ai is evolving rapidly, so it is worth staying up to all the last developments.

Some documents that I recommend that you read are:

You can find a complete list here.

Conclusion

Breaking in ai/ml may seem overwhelming, but it's about step at the same time.

• Learn the basic concepts as Python, mathematics and structures and data algorithms.
• Get your ai/ML learning of supervised learning learning, neuronal networks and transformers.
• Learn to implement ai algorithms.

The space is Ginmous, so it will probably have been completely understanding everything in this roadmap, and that is fine. There are literally degree titles dedicated to this space, which have been three years,

Just go to your own rhythm and, finally, you will get where you want to be.

Happy learning!

Another thing!

Join my free bulletin, Distribute the datawhere I share weekly advice, ideas and advice of my experience as a scientist of exercise data. In addition, as a subscriber, you will get my Free data science curriculum template!

Shoot the data | Egor Howell | Substitution
Tips and learning about data science, technology and entrepreneurship. Click to read the data, by Egor Howell, A …Newsletter.gorhowell.com