Certificate in Advanced MS Excel
Coupon code: ADVANCEXL | Offer price: 3840/-
Home > Blogs > Six Different Branches Of Specialization In AI And Which Is The Best For You?
Head spinning? Mine did too when I looked at the above chart. Don't worry, we aren’t going to cover everything on that here, nor am I going to load you with an exhaustive list of study materials. No recommendations of Andrew Ng’s courses either, I know you’re fed up of trying to break your head on Octave exercises.
Artificial Intelligence has identifiable roots in a number of older disciplines, particularly:
4. Psychology/Cognitive Science
You must have come across this meme: Statistics hiding behind the mask of ML which in turn is hiding behind the umbrella term AI. It’s true to some extent, but there’s quite some difference between a statistician and an AI researcher.
So how does one get started with AI? By studying the overlapping concepts, e.g. between philosophy and logic, or between mathematics and computation? Yes, but it’s time-consuming.
So, the absolute picks for you are these six branches:
1. Machine Learning
Were you a child who has played many guessing games, and made your friends forcefully guess what’s it that you were going to say? Then, you already have a machine learning mind.
The race is on to invent the ultimate one-size-fits-all algorithm: one capable of discovering any knowledge from data, which Pedro Domingos calls “The Master Algorithm”. He charts a course through machine learning’s five major schools of thought, showing how they turn ideas from neuroscience, evolution, psychology, physics and statistics into off-the-rack algorithms.
Skills required: Statistics and Probability, Linear Algebra, R or Python language, ML algorithms overview
a) KNIME, the Konstanz Information Miner, is a data analytics, reporting and integration platform.
b) Apache Mahout provides distributed ML algorithms focused primarily in the areas of collaborative filtering, clustering and classification in Hadoop platform.
c) Python libraries: scikit-learn, matplotlib.
b) Invest in further studies, get certified in ML and watch how it paves the way for your AI/ML career.
c) Listening to an episode a day of the podcast “Learning Machines 101” will give you an edge over other ML learners.
2. Data Science
Data-ism. If you could relate to this term, welcome to the world of data science.
For those who are new to that term, Data-ism is “an ideology that believes that liberating the flow of data is the supreme value of the universe and that it could be the key to unleashing the greatest scientific revolution in the history of humanity”, writes Yuval Noah Harari in Homo Deus.
Skills required: All the skills needed for ML plus, Big data tools like Apache Pig, Spark, Kafka, Tableau, Advanced MS Excel
a) RapidMiner is a data science software platform that provides an integrated environment for data preparation, machine learning, deep learning, text mining, and predictive analytics.
b) The H2O software runs can be called from the statistical package R, Python, and other environments. It is used for exploring and analyzing datasets.
c) Python libraries: pandas, numpy, scikit-learn, matplotlib
a) Working professionals out there, it’s time to put another feather in your cap by enrolling yourself into this Data Science PG Diploma Course.
3. Natural Language Processing
Ever been in awe with word clouds, those magical visualizations? If yes, you'll enjoy learning Natural Language Processing, NLP, in short.
a) NLTK, is your holy grail library that is used in NLP. Master all the modules in it and you’ll be a pro text analyzer in no time. Excited enough? Never mind!
b) Other python libraries: pandas, numpy, textblob, matplotlib, wordcloud
Here's a visualisation of the text from The Gospel of Buddha as word cloud image by a Kaggler, Benjamin Taylor.
a) Learn to play with the humongous amount of text using “Natural Language Processing with Python”, their official online material that provides an introduction to NLP using the Python and its NLTK library.
b) Create some breathtaking visualizations by taking learnings from Data Visualisation for Data Science using R
c) Catch up on the podcast “The Data Skeptic” by the couple Kyle and Linh Da, along with their pet parrot Yoshi who know how to make analytics a fun learning.
4. Computer Vision
Have you tried learning a new language by labeling the objects in your room with the native language and translated words? It seems to be an effective vocab builder since you see the words over and over again.
Same is the case with the computers powered with computer vision. They learn by labeling or classifying different objects that they come across and grasp the meanings or interpret, but at a much faster pace than humans (like those robots in sci-fi movies).
1. Using traditional Machine Learning Algorithms: Classical Computer Vision.
2. Using Deep Learning Libraries: Modern Computer Vision
If you are among the ones who try out every Snapchat filter ever created, the gender swap being the latest one, and gets bored of it easily, it’s high time you start making a new one yourself, programmatically! How? Computer vision knows the answer, play with it a little and it’ll spill the pixels, er, I mean beans.
But before that, you need to know about Image Processing and Fuzzification (required skills)
1. The tool OpenCV enables processing of images by applying on them mathematical operations. Remember that elective subject in engineering days called “Fuzzy Logic”? Yes, that approach is used in Image processing that makes it much easier for computer vision engineers to fuzzify or blur the readings that can’t be put in a crisp Yes/No or True/False category.
2. OpenTLA is used for video tracking which is the process to locate a moving object(s) using a camera video stream.
1. Get a new robotic vision on completing NPTEL's course on Computer vision.
2. “Computer Vision: Algorithms and Applications” is a quintessential book which deserves a mandatory read.
5. Deep Learning
Deep learning (DL) is a branch of Machine learning and its algorithms are applied in successive layers where each layer uses the output of the previous one.
Data-ism is spreading like wildfire and Deep learning is helping the AI/ML/DL community to process the huge amount of data in the speed that they’re getting ingested into the systems.
Skills required: All the skills needed for ML plus, Neural Networks overview, cognitive science, psychology
1. TensorFlow: A library for numerical computation using data flow graphs, developed for the purposes of conducting machine learning and deep neural networks research.
2. Other python libraries: Pytorch, Theano, Caffe, Keras
1. Enroll yourself into this Deep Learning course for fun-filled, immersive learning.
2. DL, DL everywhere, not a great book to read… Oh wait, here’s one: Hands-On Machine Learning with Scikit-Learn and TensorFlow (2nd edition).
3. A weekly dose of AI for your ears: “This Week in Machine Learning & AI”.
6. Deep Reinforcement Learning
Deep reinforcement learning encompasses many versatile tools for designing learning agents that can perform well on a variety of high-dimensional visual tasks, ranging from video games to robotic manipulation.
Reinforcement learning (RL) provides a powerful framework for learning behavior from high-level goals. RL has been combined with deep networks to learn policies for problems such as Atari games.
Deep Q-networks are one way to solve the deep RL problem. The idea is to execute many instances of our agent in parallel but using a shared model. This provides a viable alternative to experience replay since parallelization also diversifies and decorrelates the data.
Known Deep Reinforcement Learning applications: Self-driving cars, Robotics
Skills required: the Same skillset as DL, plus Familiarity with Reinforcement Learning terms like agent, action, environment, state, reward, policy, etc.
1. DeepMind Lab provides a suite of challenging 3D navigation and puzzle-solving tasks for learning agents. Its primary purpose is to act as a testbed for research, especially in deep reinforcement learning.
2. Simulink and Simscape Environments: Use Simulink and Simscape models to represent an environment. Specify the observation, action, and reward signals within the model.
3. GPU Acceleration: Speed up deep neural network training and inference with high-performance NVIDIA GPUs.
1. Stanford online learning never disappoints. Reinforcement Learning Winter 2019
2. Why not try something light? Like lecture slides by UC Berkeley Robot Learning Lab
And that's about it. I’ve put some light on various AI specializations and provided sufficient suggestions for you to decide which career path in AI suits you the best. Do check out our Manipal Pro Learn courses to upskill yourself in AI, ML, DL, & Data Science.