Basics Deep Learning Interview Questions you must know.
(At the end of the Blog post there are two Deep Learning project that you can do to make your CV even Stronger)
Q1. Differentiate between Artificial Intelligence(AI), Machine Learning and Deep Learning.
Artificial Intelligence is just a technique which enables machines to mimic human behavior.
Machine Learning is a part of AI technique which uses statistical methods allow machines to boost with experience.
Deep learning is a part of ML which can make the computation of multi-layer neural network feasible. It uses Neural networks to simulate human-like higher cognitive process. More detailed explanation
Q2. You think Deep Learning is Much better than Machine Learning? If that's the case, why?
Though traditional ML algorithms solve plenty of our cases, they're not useful while working together with high dimensional data, that's where we have got a sizable quantity of inputs and outputs. As an example, in the case of handwriting recognition model, we've got a wide range of input where we can have an alternative form of inputs associated with various form of handwriting.
The 2nd major challenge with traditional machine learning is to manually tell computers about the features to be used in predicting the result with better accuracy.
Q3. What's Perceptron? And So how exactly does it Work?
When we concentrate on the structure of a biological neuron, it's dendrites which are accustomed to receive inputs. These inputs are summed in the cell body and utilising the Axon it's offered to another biological neuron as shown below.
- Dendrite: Receives signals from other neurons
- Cell Body: Sums all of the inputs
- Axon: It's used to transmit signals to one other cells
Similarly, a perceptron receives multiple inputs, applies various transformations and functions and gives an output. A Perceptron is just a linear model useful for binary classification. It models a neuron that includes a group of inputs, every one of which will be given a certain weight. The neuron computes some function on these weighted inputs and provides output.
Q4. What's the role of weights and bias?
For a perceptron, there may be an additional input called bias.Whilst the weights determine the slope of the classifier line, using bias we can shift the line towards left or right. Normally bias is treated as another weighted input having the input price x0.
Q5. What is the role of activation functions?give some exapmles of activation functions!
Activation functions are used to translates the inputs into outputs. Activation function decides whether a neuron will be activated or not by calculating the weighted sum and further adding bias with it. The main objective of the activation is to introduce non-linearity to the output of a neuron for better learning.
There are many Activation functions like:
- Linear or Identity
- Unit or Binary Step
- Sigmoid or Logistic
- ReLU(Most used)
Q6.What Are the Different Layers on CNN(Convolutional Nueural Network)?
There are four layers in CNN:
- Convolutional Layer - the layer that performs a convolutional operation(applied on the input data using a convolution filter to produce a feature map), creating several smaller picture windows to go over the data.
- ReLU Layer - it brings non-linearity to the network and converts all the negative pixels to zero. The final feature maps are not the sums, but the ReLU function applied to them.
- Pooling Layer - pooling is a down-sampling operation used to reduce the dimensionality of the feature map.
- Fully Connected Layer - this layer recognizes and classifies the objects in the image, After the convolution + pooling layers we add few fully connected layers to wrap up the CNN architecture.
This post will be updated regularly and more new questions will be added