Machine Learning 1
Machine Learning : if you talk about artificial intelligence , then Machine learning is one way to feed artificial intelligence to obtain intelligence in making decision . Machine learning use statistics ,algebra ,probability and other mathematical related algorithm.
so there is always question ,what we should do to use machine learning in daily life ,because if any where you are working ,you can use AI ,but for that you should know Machine Learning ,Deep learning and Python programming and logic to fit all these in a pattern .
so there is always question ,what we should do to use machine learning in daily life ,because if any where you are working ,you can use AI ,but for that you should know Machine Learning ,Deep learning and Python programming and logic to fit all these in a pattern .
so first we will go with machine learning basic concept :
machine learning use training data for making algorithm to predict future decision or daily routine decision . enough past data is feed into training set to make algorithm.
Machine Learning information can be divided into following categories :
machine learning use training data for making algorithm to predict future decision or daily routine decision . enough past data is feed into training set to make algorithm.
Machine Learning information can be divided into following categories :
1: Unsupervised learning .
2: Supervised learning
3: Semi Supervised learning
Unsupervised learning : When learning data contains only indicative ,it is not complete clear and you have to use and find out the hidden information by using it .data used in unsupervised learning is called unlabelled data . this learning mostly used in finding anomalies ,hidden spam ,fraud attack in mail .
Supervised Learning: Clear input with descriptive information ,clear output is known ,then we use supervised learning . in supervised learning labelled data is used . Speech recognition ,face recognition is example of supervised learning .
supervised learning further divided into two parts:
1: Regression: 2: Classification
Regression used when you use quantitative values two analyse any pattern . price of house .
Classification is used to analyse positive and negative impact .
Semi supervised learning : it use labelled and unlabelled data for analysis purpose . when labelled data or labelled data is very costly for analysis purpose then we use semi concept .
Reinforcement learning : when learning data provides you feedback and based on feedback actions are design ,for example self driving car .
Machine Learning Algorithms :
there are lot of machine learning algorithm ,but mainly we use four type of machine learning algorithms
1:Statistics based algorithms : mostly eCommerce ,mathematical query solve
2:Artificial neural based algorithm : analysing complex data ,input and output
3:Logic based algorithm :
4:genetic based algorithm : used by biology to analysed genes ,disease that can be cure by
analysing genes .
Over fitting and Under fitting of Data:
Over fitting of data means , when you are considering much more training set for making algorithms ,so means you are feeding data that is not used by algorithms ,so it will produce some abnormal result .
while under fitting means just opposite to over fitting means you are feeding enough data to your algorithm to analyse problem this is called as under fitting of data .
Image has taken from another source but this will understand you very well .
How to avoid overfiiting:
1: use cross validation technique
2: Training data should be much smart .
3: Remove Extra features
4: Iteration upto only stop point .
5: Make simpler your data simpler by using regularization
6: using ensemble
supervised learning further divided into two parts:
1: Regression: 2: Classification
Regression used when you use quantitative values two analyse any pattern . price of house .
Classification is used to analyse positive and negative impact .
Semi supervised learning : it use labelled and unlabelled data for analysis purpose . when labelled data or labelled data is very costly for analysis purpose then we use semi concept .
Reinforcement learning : when learning data provides you feedback and based on feedback actions are design ,for example self driving car .
Machine Learning Algorithms :
there are lot of machine learning algorithm ,but mainly we use four type of machine learning algorithms
1:Statistics based algorithms : mostly eCommerce ,mathematical query solve
2:Artificial neural based algorithm : analysing complex data ,input and output
3:Logic based algorithm :
4:genetic based algorithm : used by biology to analysed genes ,disease that can be cure by
analysing genes .
Over fitting and Under fitting of Data:
Over fitting of data means , when you are considering much more training set for making algorithms ,so means you are feeding data that is not used by algorithms ,so it will produce some abnormal result .
while under fitting means just opposite to over fitting means you are feeding enough data to your algorithm to analyse problem this is called as under fitting of data .
Image has taken from another source but this will understand you very well .
How to avoid overfiiting:
1: use cross validation technique
2: Training data should be much smart .
3: Remove Extra features
4: Iteration upto only stop point .
5: Make simpler your data simpler by using regularization
6: using ensemble
Comments
Post a Comment