• ## Scientists Using AI To Recreate Smells From 500 to 100 Years Ago

### At a conference held at Dartmouth in 1956, a group of 10 scientists established the field of artificial intelligence and first introduced the term ‘artificial intelligence’.

#### 1-1. machine learning

It performs functions through field classification and clustering to develop algorithms and technologies that allow computers to learn with

Supervised Learning: Guess the value you want to predict from the training data

Self-learning: Computers learn data on their own without people.

Reinforcement Learning: In a given environment, the agent recognizes the current situation and takes an action, and receives positive and negative rewards accordingly.

#### 1-2. artificial neural network

An algorithm that has problem-solving ability by changing the strength of synaptic bonding through learning in which artificial neurons that form a network through synaptic bonding

#### 1-3. deep learning

Stacking various kinds of artificial neural networks ‘complex’ and ‘very deep’

Among the algorithms related to artificial intelligence, the most common and widely used is pattern recognition.

### 2. Pattern Recognition

#### 2-1. pattern recognition

What should be preceded in artificial intelligence: recognizing patterns for given data

usage example:

Ø Recognizing license plates at highway tollgates

Ø Diagnose a patient from complex test results

Ø Automatically correct spelling when typing on keyboard on your phone

Ø Decide which ads to show to specific users on your portal

Algorithm Used:

Ø Neighbor Classifier Algorithm

Ø Decision Tree Algorithm

Ø Artificial Neural Network Algorithm

#### 2-2. Nearest Neighbor Classifier Algorithm

· Algorithm for recognizing letters and numbers written by human hands or on car license plates

· Accuracy averages 97%.

Principle:

1-Each of the various shapes of a specific number or character is stored in units of pixels.

At this time, the more various shapes are stored, the higher the accuracy.

2-Enter a specific number or letter

3-Each pixel of the received number and one stored data are selected and compared in units of pixels.

4-If the comparison result is the same, set each pixel to white, otherwise set each pixel to black.

5-Finally, if the same thing (white) is 94%, it is judged to be a correct number, and if the same thing is 70% or less, it is judged not to match the input number.

#### 2-3. kNearest Neighbor Classification Algorithm

Analysis of where a particular voter will vote in an election.

& frequently used in various marketing analysis