Is the Deep Learning Era really coming to an end? If somebody had procrastinated this change way back in 2011 that this was going to be a hot topic for debate, the tech world would have been astonished, with comments like Wow!! You are smoking something really strong!
Almost everything we witness into artificial intelligence today is thanks to deep learning. The deep learning algorithms work by deploying statistics to find patterns in data and have proved immensely powerful in mimicking human skills such as the ability to hear and see. To a very narrow extent, deep learning algorithms can even emulate our ability to reason and comprehend.
Deep learning is a technology that powers Facebook’s news feed, Netflix’s recommendation engine and Google’s search is radically transforming industries like health care and education. But going by the MIT Tech research, machine learning that had started to pick up over the last 20 years marking a rapid growth since about 2008 is now witnessing a downfall in the research fervour which currently seems to be dying down.
Is it Really the End of an Era?
Artificial intelligence developers may soon be caught on the brink of a paradigm shift and may witness the way out of deep learning technologies, that has dominated the field for several years. This change is marked by a shift towards machine learning technologies during the late 1990s and the early 2000s marked with a rise in the popularity of neural networks that begun in the early 2010s, and the growth in reinforcement learning over the last few years.
A team from MIT Technology Review scanned all the 16,625 research papers in the artificial intelligence section of arXiv, an open-source repository for sharing research published between 1993 and November 2018. The source arXiv offers a great input for gleaning some of the larger research trends and for catapulting the push and pull of different ideas. These ideas toggle between the growth and abysses of Machined Learning to the dominance and fading out of Neural Networks to the current favorite Reinforcement learning algorithms.
The Machine Learning Transition
The biggest shift was marked in the early 2000s with the transition away from knowledge-based systems. The computer programs in the early 2000s were based on rules to encode all human knowledge. This feature could not sustain for long and, researchers turned their interest to machine learning, the parent category of algorithms that includes deep learning.
The Neural Network Dominance
Under the new machine-learning paradigm, the tectonic shift to deep learning did not happen immediately. Instead, the move tested with a variety of methods in addition to neural networks which is the core machinery of deep learning. Some of the other popular techniques in the race included evolutionary algorithms, Bayesian networks and support vector machines. All these techniques take different approaches to find patterns in data for intelligent analysis.
The Reinforcement Learning Era
Reinforcement learning (RL) is the latest buzzword in the technology evolution. RL is a component of Machine Learning and takes suitable action to maximize reward in a particular situation. RL algorithms are employed by various software and machines to find the best possible behavior or path which should be taken.
When we talk about this technology, there is a stark difference to the similar framework of supervised learning, in which the training data has the answer key to train the model with the correct answer itself. Reinforcement learning offers no-answers but the reinforcement agent plans and decides what to perform next in the given task. In the absence of a training dataset, it is bound to learn from its experience.
The Increased Adaptability of Reinforcement learning (RL)
Reinforcement learning (RL) has been deployed for long into industry use. It is used in robotics for industrial automation and into machine learning and data processing. Reinforcement learning can be used to create training systems offering customized instructions and inputs according to the requirement of the users.
Change is the only constant is a long proved adage, which is proven by the sudden rise and fall of different techniques which mark the growth of disruptive technologies for a long time. Every decade has witnessed the growth and the end of different ideas which have seen dominance. Are you ready to embrace and accept Reinforcement learning as the next age technology?
Watch this space for more!