Deep Learning is a subset of Machine Learning that has picked up in recent years.The learning comes into the picture.Some features from the object that we see around us or what we hear and various such things. Finding features is a pain-staking process. We tried to learn ,we tried to train the machine learning model which could gather information of the object from these features.
After training we give similar such features from an unknown object ,the model could identify what that unknown object is. When it comes to deep learning let the machine understand what features to look for to identify different objects or to identify different animals and the output of the model will be the decision that for which the machine or the model has been trained.
Applications of Deep Learning:
Deep learning has a major success in lots of applications and solved problems, like image recognition and speech recognition, which were consider hard for AI systems. The best example is AlphaGo Zero which uses deep learning and reinforcement learning to train an AI agent in the game of Go from scratch and beat the world champion.
The game of Go was considered unlearnable by AI due to its million possible outcomes. We will decipher more on why deep learning works in the latter part of the journey.
Deciding if ML is Needed
Before we even begin to map the journey of Data Science, we have to ask the pertinent question. do we need to apply ML to solve our problem? The flowchart shown is a handy guide to decide if the use case needs Machine Learning or not.
Most of the time, we don’t realize, but defining the right problem is extremely difficult. The judging of whether Data Science can be applied to a problem is also a quality of a Data Scientist.
Things to Ponder:
Can ML solve your business problem
• The rule-based approach gives you good accuracy, but ML gives you a marginal improvement.
so, a decision needs to be made if an ML model is worth investing in or not. Most of the time, business costs dictate whether ML is applied or not.
• Do you require prediction in your problem or causal inference?
however, If the business problem requires investigation of its causes, then ML is not a good fit.
thus, ML is good only at prediction given a host of factors.
- Is the problem self-contained and insulated by outside influences?
Model of deep learning is call as generative model. To train such generative model. we must need an adversary that can tell the generator what you have generated does not look like the natural image of the natural video.
So there is an error mechanism. error is feed back to the generative model so that it improves itself to learns a better representation of the natural image or video. however, this generative model with adversary is call as adversarial network.
From Data to Insights
This will be a typical workflow for any ML problem to generate insights from data. Let us take an example of real data and go through the different phases. During the course of the complete program, we will take up each of these phases in detail. We have abstracted the relevant phases into 4 major ones. These will help you understand the overview before we take a deep-dive into each of them in further detail in future concepts.
Step 1: Defining the Problem
Problem definitions is an often underestimate step in the entire process, but a very crucial step to success. It is important to define the scope of the problem well.
Step 2: Preparing Data
From the vast population of available data, find what are the missing values, decide which columns are relevant for your processing, and remove the rest. Find and engineer new features that you think might help in solving your problem better.
Step 3: Developing the Model
therefore, This is the part which every scientist fantasizes to actually do. hence, Develop the model on data and get the insights to answer the business questions define earlier.
Step 4: Evaluating the Model
it’s important to assess the model and see if it is working on some data before it’s deploy on data. How to assess the model and choose the best model is also important.
written by: Thomas Vengazhiyil Alex
reviewed by: Kothakota Viswanadh
If you are Interested In Machine Learning You Can Check Machine Learning Internship Program
Also Check Other Technical And Non Technical Internship Programs
