What is Artificial Intelligence (AI)?

Artificial Intelligence

What is Artificial Intelligence (AI)?

Guest Post By-  Tanya Khare

What You Need To Know About Artificial Intelligence: As we all have come to know that technology of artificial intelligence is flourishing and is a great invention. Earlier the security researchers were interested in making computers or technology that could work like humans, but then they realized that there is need of system that could not only work like humans but also update itself over time and adapt new capabilities. Therefore, they worked on creation of intelligent machines, that can pursue intellect just like human intellect.

Artificial intelligence is “Area of computer science that emphasizes the creation of intelligent machines that work and react like humans such as learning, planning, speech recognition, thinking, decision making and problem solving.”

Therefore, Artificial intelligence(AI) is also referred to as machine intelligence. The AI senses its environment and functions accordingly in order to accomplish its targets. The targets can be simple, complex and may mutate according to requirement. AI uses is a set of unambiguous instructions that a mechanical computer can execute called Algorithm. These algorithm enables the machines to learn from data and other strategies. These algorithms are divided into 3 categories: supervised learning, unsupervised learning, and reinforcement learning.

Supervised learning is the category in which the dataset is available for some labels but is needed to be predicted for other new instances. Unsupervised learning is the category in which it is difficult to ascertain implied relationships for the unlabelled dataset. Reinforcement learning is kind of dynamic programming that learns by interacting with its environment.

The algorithms under supervised learning are:-

  1. Naïve Bayes Classification: – It is based on applying Bayes’ theorem with strong (naive) independence assumptions between the features

                        P(A | B) = (P(B | A) P(A)) /   P(B)

                Where P(A | B) means  is posterior probability,

                P(B|A) is likelihood,

                P(A) is class prior probability,

                P(B) is predictor prior probability.

  1. Decision tree: – It is a support tool which uses tree graph model to make decisions and their possible consequences, including chance-event outcomes.
  2. OLSR (Ordinary Least Square Regression): – Least square method is used to perform linear regression where, Linear refers the kind of model you are using to fit the data, while least square refer to the kind of error metric you are minimizing over. It enables the line to fit into through the multiple points.
  3. Logistic regression: – It is also called log odds or logit. logical regression is described as a mathematical model to estimate the probability of an event occurring with the help of earlier information or data. Dataset in logistic regression is in binary form where the event may or may not happen under the probability of 1 or 0 .
  4. SVM (Support vector Machines): – the biggest problems that have been solved using SVMs (with suitably modified implementations) are display advertising, human splice site recognition, image-based gender detection, large-scale image classification.

Some algorithms of the unsupervised learning are: –

  1. Clustering Algorithm: – This algorithm enables the machine to group the target into similar and different groups. Some of the good clustering algorithms involve Probabilistic, Dimensionality Reduction, Neural networks / Deep Learning, etc
  2. Principle Component Analysis: – This algorithm enables the correlation of observations variables to the unrelated variables called principle components.
  3. Singular value decomposer: – It is the first face recognition algorithm and represent faces as linear combination.
  4. Independent component Analysis: – It defines a generative model for the observed multiple variety of data, which is typically given as a large database of samples. Data is assumed to be a mixer of known and unknown variable data.


What You Need To Know About Artificial Intelligence

AI works on some approaches like classic approach, which focus individual human behaviour, knowledge and inference method. On the other hand, Distributed Artificial Intelligence(DAI) focus on social behaviour, interaction and knowledge sharing among different agents. The concept of classic AI and DAI lead to an intelligent multi-agent technology

Computational Intelligence (CI) is another AI approach that sprouted AI in which the major emphasis is placed on heuristic algorithms such as fuzzy systems, neural networks, evolutionary computation, machine learning and Artificial Immune System(AIS).

AIS ( Artificial Immune System) is a nature inspired system which is adaptable to dynamic environment and capable of continuous learning by upholding stability. AIS detection comprises evolution of self-tolerance, clone, variation and antigens detection. It supports efficient spam detection with very low false positive and negative rates. AIS supports hybrid learning algorithms for anomaly detection. It promotes self-recovery and is also scalable and robust.

ANN (Artificial Neural Network) consist of artificial neurons which have the ability to learn, process, distribute information, self-organization and adaptability for solving problems. Intrusion detection system using neural network is capable of detecting all intrusion attempts without any false alerts. It is used in spam filtering using keywords. It detects potentially malicious traffic also it is very effective in detection of DDoS.

Genetic Algorithms and Fuzzy Sets Applications are learning algorithms for anomaly detection. Fuzzy Logic (FL) is a method of reasoning that resembles human reasoning. FL is the approach that emulate the way of decision making in humans and consider all intermediate possibilities between digital values, fuzzy host-based IDS uses data mining techniques. Confidentiality, Integrity and availability are improved through genetic algorithm Rule-based IDS. FL are used to classify network attacks data, while genetic algorithm optimizes the findings through fuzzy rule. The detection of network takes place in real time. Its detects packet dropping attacks with high positive rates.