MLearning.ai
Published in

MLearning.ai

TERMINOLOGIES TO KNOW AS A MACHINE LEARNING ENGINEER — PART 2

In the last blog, we have discussed the term Machine learning, and it paradigms. Here, we’ll look into some basics terms that a machine learning engineer or a data science specialist should be aware of.

Machine learning, image from iStock

Normalization

Machine can understand the data only in the form of numbers. Even if the given data are images, audio or video, the data is converted into numerical values before it is fed into the model. But the converted numerical values of various ranges i.e. the number can be of any range, which can led to complex process and increase in latency. Hence, normalization is carried out.

Normalization is the process of converting the higher values to the values between [0,1]. This makes the computation simple for the machine.

Generalization

Generalization is a term which refers to the adaptability of the model to a new unseen data and produce the output. It is said that if the model is too trained in the training data, then it will make an inaccurate predictions on the new data that is fed to the model, making the model useless.

Model fitting

Model fitting is the measure of how well the machine learning model generalizes to similar data that was used for training the model.

Overfitting

If the model makes an inaccurate predictions when given a new data, even though it can able to make accurate predictions for the training data is called overfitting. In other words, the model has good performance in training data but has poor generalization to other data.

Underfitting

It can be defined as the performance of the model will be poor with training data as well as poor generalization to other data.

Outliers

In a dataset, there will be some points that are significantly different from the rest of the data points. Such points are called outliers. The outliers lie at a separate region from other data points which lies as a group.

Reasons for outliers

  • The outliers occur due to errors during the data entry or occurance of any fault during measuring the data
  • Else it can occur due to natural occurence such as the salary data of junior employee will be very much varying when compared to senior employee
Outliers, image from Data science foundation

Downsampling

Downsampling in image is basically known as reducing the resolution while maintaining the same two-dimensional (2D) resolution. Downsampling is carried out in order to reduce the storage required. The image is converted into vectors i.e. matrix with numerical values then they are downsampled based on the frames size that has been chosen.

Downsampling is needed because to create small machine learning models that does not require many training data points to predict the output.

Back propagation

Back propagation is the ability of the model to come back from the forward layer to the previous layer to update the weights of the neurons. The weights are updated to improve accuracy of the model, so that the desired output is achieved.

Accuracy

Accuracy of a machine learning model describes how good is the performance of the model. If the accuracy is high, then the model has good performance. Accuracy can be defined as the ratio of correct predictions to the total number of predictions.

Precision

Precision of a model is to define that how quality the predictions be, means that the clarity or how exact the predictions are made.

Recall

It is the ability of the model to find all the relevant cases within a dataset. This measures the ability of model to detect the positive samples.

F-score

F-score is also used to determine the performance of a model. F-score is calculated by combining the precision and recall into a single measure. Once the precision and recall is calculated for a classification problem, then the F- score is calculated by combining those by the following formula,

F-Measure = (2 * Precision * Recall) / (Precision + Recall)

So far we have covered the most common basic terms to get started with Machine learning. In the upcoming blogs, we’ll dive into some basic processes and about models in Machine learning.

Happy learning!!

--

--

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Sudarshan S

Sudarshan S

72 Followers

Tech enthusiast | Developer | Programmer | Cybersecurity | Machine learning | Data science