Original article was published on Deep Learning on Medium
What’s artificial intelligence (AI):
It’s a level of intelligence demonstrated by the machine, where it tries to mimic the actions with the level of the human’s, which is pretty ironic when we say that computers are more intelligent than humans.
In normal programming we feed the machine with the information, the type of information to treat and the type of output to expect as result, basically, any mean to think or to solve a problem we think about it first and then the machine all it has to do is take the input to give an output base on the conditions we specified, it’s fast that’s for sure but not intelligent.
A code where it can solve any problem isn’t an intelligent code, it just happens that the coder has thought of every detail to solve the problem, including the special and edge cases, a phenomenon is known as the AI effect.
AI was founded in 1955, after many ups and downs. It’s so big of a concept that it has many sub-fields such as Machine Learning, Neural Networks, Evolutionary Computation, Vision, Robotics, Expert Systems, Speech Processing, Natural Language Processing, and Planning.
To be capable of understanding AI, first, you need to understand the fields that it’s based on like computer science, information engineering, mathematics, psychology, linguistics, philosophy, and many others, and it uses many tools including search and mathematical optimization, artificial neural networks, and methods based on statistics, probability, and economics.
It was created based on the idea that machines can take over our tasks making free and on the vacation, all year round, wich also for the same reason many people consider it a danger for humanity and create a risk of mass unemployment.
Nowadays many resources have been invested to develop this technology hoping to solve many challenging problems in computer science, software engineering and operations research.
What’s machine learning (ML):
It is the study of computer algorithms that improve automatically through experience, it’s a sub-fieald bases on AI. It’s based on mathematical model, where it can solve problems that are impossible to solve by any normel means.
It’s also a result of group of many fields like computational statistics, mathematical optimization, Data mining, exploratory data analysis, and unsupervised learning.
ML algorithms are used in many applications such as :
- Email Spam and Malware Filtering
- Online Customer Support
- Search Engine Result Refining
- Product Recommendations
- Online Fraud Detection
- Social Media Services
- Videos Surveillance
- Predictions while Commuting
But what’s the difference between AI and ML? well, AI as mentioned above it require to be walked through the problem step by step tell you to find a solution, but in ML all it needs is time and resources, it involves computers discovering how they can perform tasks without being explicitly programmed to do so. It involves computers learning from data provided so that they carry out certain tasks.
For simple task assigned to computers, it simpler to write algorithms step by step guiding the machine to solve the task, but when the problem needed to solve is much bigger it will be a bit hard to impossible to define all the edge cases possible with a solution to treat them because that’s way above the human brain capability, and that’s where machine learning come in handy, it can turn out to be more effective to help the machine develop its own algorithm, better than have human programmers specify every needed step.
“At its heart, machine learning is the task of making computers more intelligent without explicitly teaching them how to behave. It does so by identifying patterns in data — especially useful for diverse, high-dimensional data such as images and patient health records.” –Bill Brock, VP of engineering at Very.
And to solve a problem the machine need a method or as called Approaches wish can be divided into three categories depending on the signal available in the learning system:
- Supervised learning
- Unsupervised learning
- Reinforcement learning
And many other approaches have been created wich doesn’t fit in any of the main ones, also if the task requires it can use more then one approach to solve it.
Supervised learning requires data of the input and the desired output, called training data which contains a set of examples to follow. A good algorithm is when it can solve a task out of the training data given perfectly fine. It can have many types like Active learning, classification, and regression.
Unlike the supervised learning, the data it uses is the unlabeled data, it’s not used to solve a problem but to explore how the machine function and how it treats the data when it’s unlabeled
It’s the trial and error methods, where it discovers solutions or errors, it’s known for being the slowest approach, simple feedback is required to learn which action is best, this is known as the reinforcement signal.
It’s somewhere in between supervised and unsupervised learning, it requires both labeled and unlabeled data to function. Usually, it only requires unlabeled data, but in case of the data requires skilled and relevant resources in order to learn from it then a labeled data is required.
Wha’s Deep Learning (DL):
Deep learning is a part of a machine learning family wich based on artificial neural networks with representation learning. It’s present in many fields such as computer vision, machine vision, speech recognition, natural language processing, audio recognition, social network filtering, machine translation, bioinformatics, drug design, medical image analysis, material inspection, and board game programs. And it’s based on deep neural networks, deep belief networks, recurrent neural networks, and convolutional neural networks.
The name deep learning was inspired by how it divides the work to multiple layers, and each layer has layers underneath it more layers.
Neural networks were inspired by information processing and distributed communication nodes in our biological systems.
Artificial neural networks are computing systems inspired by the biological neural networks that constitute animal brains. Such systems learn (progressively improve their ability) to do tasks by considering examples, generally without task-specific programming. For example, in image recognition, they might learn to identify images that contain cats by analyzing example images that have been manually labeled as “cat” or “no cat” and using the analytic results to identify cats in other images. They have found most use in applications difficult to express with a traditional computer algorithm using rule-based programming.
Typically, neurons are organized in layers. Different layers may perform different kinds of transformations on their inputs. Signals travel from the first input, to the last output layer, possibly after traversing the layers multiple times. The original goal of the neural network approach was to solve problems in the same way that a human brain would. Over time, attention focused on matching specific mental abilities, leading to deviations from biology such as backpropagation, or passing information in the reverse direction and adjusting the network to reflect that information.