Machine Learning Algorithm
Modern hair coloring relates to a chemical process that either add or remove pigments from hair.
So whenever starting the hair coloring you should keep in mind the contribution of a pigment of your hair. From three different shades of color, many brunettes and blondes have the same contributing pigment, it’s just varying only by the lightness and strength of the color.
Another important thing to determine in hair coloring is Tone, which indicates up to which level the hair color is shown its warmness or coolness.
After seeing lots of things before hair coloring you have to need to give a final look towards base color. As today’s hair color formulas clearly uploaded base color on the bottle. So, for careful evaluation of hair coloring, you should try a customization of your color requirement specifically relates with your picture. For this Kiran is tried to assessed use of machine learning to develop an algorithm which helps to assign colors and tells the machine the exact amount needed for base color to added to the final mix.
Categories of Hair Color assigned to machine:
Here are some important ML algorithms, through which following some rules helps to assign the color, amount of base color to add to the final mix.
• Supervised Learning: In Supervised learning system labeled training data used to learn the mapping function from input variable (X) to the output variable (Y).
Y = f(X)
Supervised Learning problems can be of two types:
? Classification- Outcome from given sample is in the form of categories.
? Regression- Outcome from the given sample is in the form of real values.
• Unsupervised Learning: Unsupervised learning problems worked only the input variable means uses unlabeled training data to model the structure of data underlying in.
Like Supervised learning UL problems also can be of two types:
? Association- Probably the co-occurrence of items in collection and is used in market-basket analysis.
? Clustering- Sample objects within same cluster are more similar as comparison o objects from another cluster.
• Reinforcement Learning: Reinforcement learning is best type of learning algorithm as it allows the agent to decide the next action that would be best and based on its current state, by learning behaviours that will maximize the value. It works like trial and error, so it basically robotics-based algorithm learning.