What is Deep Learning ?
Have you ever thought about how autonomous cars are even possible?
How can virtual assistants understand our problems and give a solution?
Have you ever wondered how text translations can be performed very accurately?
Al of these and more are produced by Deep Learning and Artificial Neural Network. Marcus who is a professor at New York University and CEO of Geometric Intelligence stated that “Deep learning has since yielded numerous state of the art results, in domains such as speech recognition, image recognition , and language translation and plays a role in a wide swath of current AI applications.” As we know from our previous post, Artificial Intelligence covers Machine Learning, and on the other hand, Deep Learning is a subset of Machine Learning. Artificial intelligence (AI) is a broad term that refers to methods that allow computers to imitate human behavior. All of this is made possible by machine learning, which is a series of algorithms trained on data.
On the other hand, deep learning is a type of machine learning technique that teaches the model using human learning activation: learn by example. That means the deep learning model learns from data and perform by these. In the process of learning, artificial neural network structure, which is inspired by the human brain structure, is used. Artificial neural networks are multi-layer neural networks that we use to identify patterns, classify variables, make predictions, and so on.
The artificial neural network architecture generally consists of several layers, and each layer has specific neurons. When the first layer is called the input layer, the last one is called the output layer. Other layers are hidden layers. The process starts with receiving the vector of input to the input layer. By assigning weights on input between layers and applying mathematical formulation, the output is obtained.
Deep learning applications
Self-driving cars
Companies try to teach computers how to control the car with digital sensor systems without human senses. To achieve it, companies train deep learning algorithms using a mass of data. The most recent example of that is Tesla Autopilot AI. The company created an Autopilot neural network which consists of 48 networks, and it can manage the car without human intervention. 70.000 GPU hours were spent for its training. Autopilot AI has abilities to the detection of items in the near, prediction of other cars’ movement, and path planning.
Deep Learning in Healthcare
Deep learning is being used by cancer researchers to identify cancer cells automatically. UCLA researchers created a high-dimensional data collection that was used to train a deep learning application to reliably classify cancer cells.
Voice Search & Voice-Activated Assistants
Both the realms of industry and academia have learned deeply to recognize speech. Xbox, Skype, Google Now, and Apple Siri already use their technologies to identify human speech and speech patterns using deep learning techniques.
Recommendation Systems
Based on previous conduct, Amazon and Netflix popularized the concept of a suggestion system that gives you an excellent opportunity to know what you may want next. In complex contexts, like music interests or dress preferences across many platforms, deep learning can be used to enhanced suggestions.
Source
Marcus, G. (2018). Deep Learning: A Critical Appraisal. CoRR, abs/1801.00631.
Education, I. C. (2020, May 1). Deep Learning. IBM. https://www.ibm.com/cloud/learn/deep-learning