In this blog, we will be discussing the difference among popular buzzwords i.e., Machine Learning (ML), Artificial Intelligence (AI), and Deep Learning (DL).
Further, we will address the superiority of Deep Learning over Machine Learning along with challenges faced while working in this field and solutions to them.
First of all, let’s look at some of the facts and figures of the impact of AI on the IT industry at large!
- 60% of videos watched on YouTube are a result of recommendations (AI).
- 40% of apps installed from PlayStore are a result of AI-driven recommendations.
- AI can accelerate business productivity by 40%.
- The number of AI-based start-ups has amplified to 14% since the 2000s.
Despite many discussions on AI, DL, and ML, a lot of misconception persists. The worst misconception being “All these terms signify the same meaning”. However, these terms are related to each other but not the same.
AI is a broader term that describes the capability of the machine to learn and solve problems just like humans. In other words, AI refers to the replication of humans, how it thinks, works and functions.
On the lower ground, AI can be seen as a program that instructs the machine to function in a certain way according to the situation. So, we can call a bunch of if-else statements, AI. For instance, self-driving car moving on the road. A small algorithmic rule for such a car can be:
if distance_from_in_front <100m: stop(); else: keep_moving();
Relation of AI with ML and DL
There are two ways of incorporating intelligence in artificial things i.e., to achieve artificial intelligence. One is through machine learning and another is through deep learning. That means DL and ML are ways of achieving AI.
Now that we have understood the term “AI”, we can take a closer look on ML and DL.
ML comprises algorithms for accomplishing different types of tasks such as classification, regression, or clustering. The accuracy of algorithms increases with an increase in data.
“Technique to learn from data through training and then apply learning to make an informed decision”
Analyzing and learning from data comes under the training part of the machine learning model. During the training of the model, the objective is to minimize the loss between actual and predicted value. For example, in the case of recommending items to a user, the objective is to minimize the difference between the predicted rating of an item by the model and the actual rating given by the user.
“Difference between the predicted and actual value is computed using loss-function or objective function. Therefore, defining the objective/loss function is the gist of ML model.”
In today’s era, ML has shown great impact on every industry ranging from weather forecasting, Netflix recommendations, stock prediction, to malware detection. ML though effective is an old field that has been in use since the 1980s and surrounds algorithms from then. Below is a small snippet of the ML model.
from sklearn.linear_model import LinearRegression predictor = LinearRegression(n_jobs=-1) predictor.fit(X=TRAIN_INPUT, y=TRAIN_OUTPUT)
Deep Learning – The Future
Deep learning is an emerging field that has been in steady use since its inception in the field in 2010. It is based on an artificial neural network which is nothing but a mimic of the working of the human brain.
Just like the ML model, the DL model requires a large amount of data to learn and make an informed decision and is therefore also considered a subset of ML. This is one of the reasons for the misconception that ML and DL are the same. However, the DL model is based on artificial neural networks which have the capability of solving tasks which ML is unable to solve.
The future is driven by DL models. Without DL, Alexa, Siri, Google Voice Assistant, Google Translation, Self-driving cars are not possible. To learn more about building DL models, have a look at my blog on Deep Learning in-depth.
Two major advantages of DL over ML:
1. Feature Extraction
Machine learning algorithms such as Naive Bayes, Logistic Regression, SVM, etc., are termed as “flat algorithms”. By flat, we mean, these algorithms require pre-processing phase (known as Feature Extraction which is quite complicated and computationally expensive) before been applied to data such as images, text, CSV. For instance, if we want to determine whether a particular image is of a cat or dog using the ML model. We have to manually extract features from the image such as size, color, shape, etc., and then give these features to the ML model to identify whether the image is of a dog or cat.
However, DL models do not any feature extraction pre-processing step and are capable of classifying data into different classes and categories themselves. That is, in the case of identification of cat or dog in the image, we do not need to extract features from the image and give it to the DL model. But, the image can be given as the direct input to the DL model whose job is then to classify it without human intervention.
Raw Data is given to DL model. Pre-processed data is given to ML model.
2. Big Data
With technology and the ever-increasing use of the web, it is estimated that every second 1.7MB of data is generated by every person on the planet Earth. Therefore, analyzing and learning from data is of utmost importance.
Deep Learning is seen as a rocket whose fuel is data.
The accuracy of ML models stops increasing with an increasing amount of data after a point while the accuracy of the DL model keeps on increasing with increasing data.
Below is a small snippet to build a DL model:
from keras.models import Sequential from keras.layers.core import Dense, Activationmodel = Sequential() model.add(Dense(10, activation=’sigmoid’, input_shape=(784,))) model.add(Dense(10, activation=’softmax’))
Challenges faced in the field of AI, ML, and DL along with its Solutions:
1. Computing Complexity
ML and DL algorithms require large data to work upon and thus need quick calculations i.e., large processing power is required. However, it came out that limited resources are available to implement these algorithms on large data.
Solution: Cloud computing such as Google Colab, Kaggle, Microsoft Azure brings some hope but as the volume of data goes up, the complexity of algorithms starts increasing and these resources could not help!
2. Lacks support and awareness
Unlike web development and software development, AI is quite a new field and therefore lacks many use-cases which make it difficult for many organizations to invest money in AI-based projects. In other words, there are comparatively fewer data scientists who can make others believe in the power of AI.
Solution: The remedy is to make people aware of the power of AI. Further, instead of building everything from scratch, enabling organizations to take ready-made solutions and just plug and play with data – AI-driven services.
3. Black-box Nature
AI-based model is black-box in nature which means all data scientists have to do is find and import the right artificial network or machine learning algorithm. However, they remain unaware of how decisions are made by the model and thus lose the trust and comfortability of data scientists.
Solution: One possible solution is making people believe that model really works well. Another possibility that is trending nowadays is “Explainable AI”, letting people know the reason behind the decision.
4. Data breach and Identity Theft
ML and DL algorithms require a large amount of data to learn and thus make informed decisions. However, data often contain sensitive and personal information which makes models susceptible to identity theft and data breach.
Solution: Nowadays, various privacy and security solutions are being provided to protect sensitive information. In addition, the implementation of the General Data Protection Regulation (GDPR) is being done by the European Union to ensure the protection of personal data.
5. Data Sparsity
It is a fact that today data generated is much greater than ever before. But still, there lack datasets with a great density that be used for testing AI algorithms. For instance, the standard dataset used for testing the AI-based recommendation system is 97% sparse.
Solution: Academic and industry researchers have now started working on developing AI models that can work on sparse data without compromising with accuracy.
Steps to be followed while Building AI model:
1. The problem should be defined with clear objectives to be achieved.
2. Identify, collect, and understand the data for testing.
3. Define the objective function.
4. Tune the hyperparameters and train the model.
5. Evaluation metrics should be fixed. Accuracy should not be the only parameter while fixing evaluation metrics.
6. Testing should be rigorously done from all perspectives.
Thanks for reading!
I hope you enjoyed learning the difference between Artificial Intelligence, Machine Learning, and Deep Learning.
If you liked this and want to know more, go visit my other articles on Data Science and Machine Learning by clicking on the Link.
The media shown in this article are not owned by Analytics Vidhya and are used at the Author’s discretion.