Machine Learning vs. Deep Learning
Over the past few year , the term deep learning and machine learning is very popular into business language when discussion is about Analytics, Big Data and Artificial Intelligence (AI).
Just for the information below is the google trend for these keywords.
These two keywords are often used in such a way that they seems like interchangeable buzzword, but there is lot of difference between them. As we move forward through the content i will try to explain the difference between them. So keep reading …
In short — Machine and Deep learning both term are related to Artificial Intelligence. Machine Learning is getting machine a learning ability to act like a human being without being explicitly programmed. Deep Learning is subgroup of machine learning. Let’s dig a little more into this.
Machine Learning and Deep learning are both part of Artificial Intelligence, with AI which came into picture first, then came the machine learning and now deep learning is flourishing and solving some of the complex real life problem. Their relationship is visualized with the help of below diagram.
Machine Learning — An Approach to Achieve Artificial Intelligence
“Field of study that gives computers the ability to learn without being explicitly programmed” — Arthur Samuel
Machine Learning is the practice of using algorithm to break up data, learn from them and then use this learning to make some prediction about certain things. We are not doing any hard-coding with some specific set of instruction to accomplice any task, instead machine is trained with huge amount of data which give an ability to trained model so that it can perform specific task, i.e. Machine-learning programs, in a sense, adjust themselves in response to the data they’re exposed to.
“A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P, if its performance at tasks in T, as measured by P, improves with experience E.” — Tom Michel
Machine learning is closely related to (and often overlaps with) computational statistics, which also focuses on prediction-making through the use of computers. It has strong ties to mathematical optimization. The “learning” part of machine learning means that ML algorithms attempt to optimize along a certain dimension; i.e. they usually try to minimize error or maximize the likelihood of their predictions being true.
Deep Learning- A Technique for implementing Machine Learning
“The analogy to deep learning is that the rocket engine is the deep learning models and the fuel is the huge amounts of data we can feed to these algorithms.” — Andrew Ng
Deep Learning is subset of Machine Learning. Deep learning model involves feeding a computer system lot of data, which it can use to make decision about other data. Deep learning works in same way as human brain make conclusion with respect to any scenario. Mainly when people uses the term deep learning, they are referring to deep artificial neural networks.
Deep artificial neural network are a set of algorithms which have sets new records in accuracy for many important problems, such as image recognition, sound recognition, recommended system, and many more. One of the famous record setup by deep learning algorithm is Deep mind well-known AlphaGo, which beats the former world champion in 2016 and 2017.
The word Deep means number of layers in a neural network. A shallow network has only one so-called hidden layer, and a deep network has more than one. Multiple hidden layer in a neural network allow to learn features of the data in a so-called feature hierarchy, because simple features (e.g. two pixels) recombine from one layer to the next, to form more complex features (e.g. a line). To learns all the featured ANN required lot of computational power, because of this now a days GPUs are high in demand for training the deep-learning model.
By comparing the Machine and Deep Learning we can say that deep learning tends to results in higher accuracy, requires more hardware power and works very well on unstructured data such as pixels, texts or blob.
Comparison of Machine Learning and Deep Learning
Now that we have got some basic idea about ML and DL. We will try to compare to techniques.
Hardware Dependencies
Deep learning requires high-end machines because while doing features extractions and classification at different part of hidden layers requires lot of large matrix multiplication, contrary to traditional machine learning algorithms, which can work on low-end machines.
Feature Extraction
In Machine Learning, most of the applied features need to be identified by an experts and then hard-coded as per the domain and data type. Eg. features can be pixels values, textures, shape, position and orientation. The performance of most of the ML algorithm depends on how accurately the features are identified and extracted.
Deep Learning algorithms try to learn high-level features from data. This is very distinctive part of deep learning and a major step ahead of traditional machine learning. Therefore, deep learning reduces the task of developing a new feature extractor of every problem. Eg. CNN (Convolutional Neural Network) will try to learn low-level features such as edge and lines at early layers and then high level features in next hidden layers.
Data Dependencies
Performance of both techniques differ as the scale of data increases. When data is less Deep learning doesn’t perform very well because of the huge number of parameters needed to be tuned by a learning algorithm. Deep learning starts with some random parameters and then some gradient based optimization algorithm is used to converge the network to an optimum solution, which might not be global optimum. This whole process requires lot of data. Deep neural networks have many false positive initially and slightly improves with every learning iteration. On the other hand Machine learning algorithm have their handcrafted rules which works in less amount of data.
Problem-Solving Approach
When solving a problem using traditional ML algorithms, it is generally recommended to break the task into different parts, solve them individually, and combine them to get results. Deep learning, by contrast, believes in solving problems end-to-end. Eg. Suppose we have to find multiple objects in an image and name them. Using ML algorithm this task is divided into two parts: object detection and object recognition. First use bounding boxes to find the objects in an image then classify the detected object using algorithm like SVM with HOG.
On the contrary, in deep learning algorithm, you would do process end-to-end.Eg. using YOLO network you would pass in an image and it would give out the location along with name of detected object.
Execution Time
Because there is lot of parameters in deep learning algorithm it requires lot of time to train them, whereas machine learning comparatively takes much less time to train. This in turns completely reverse on testing time. At test time, deep learning algorithm takes much less time to run.
Interpretability
Deep learning somewhat behaves like a black box means we don’t know what the neurons were supposed to model and what these layers of neurons were doing collectively. So we fails to interpret the result.
On the other hand, machine learning algorithm like decision tree give us crisp rules as to why they chose what they chose, so it is particularly easy to interpret the reasoning behind it.
Uses of Machine learning and Deep Learning
These algorithms have vast applications. Below few are taken from Wikipedia.
— Computer Vision: Used for facial recognition and vehicle plate detection.
— Medical diagnosis: Used for Cancer detection and many more anomaly detection.
— Natural Language Processing: Used for sentiment analysis.
— Information retrieval: Eg. Search Engines
So that’s all for this post. Hopefully you liked this post :)