Posted on

5 AI Trends to Watch in 2018 (via O’Reilly Media)

ai, trends, machine learning, deep learning, learning

Introduction

From methods to tools to ethics, Ben Lorica looks at what’s in store for artificial intelligence.  Here are five trends to look for in 2018.

Source: O’Reilly Media

 

5 AI Trends for 2018

  1. Progress in machine learning processes – Recently deep learning architectures and algorithms are having vast improvements in the areas of computer vision, speech and text.  Companies expect to use better algorithms.  Because of this, they are implementing training, inference and data processing on newer edge devices.  Machine learning experts will work on new ways to analyze more data through neuro-evolution and deep learning.   With how technical and scientific machine learning is becoming, people are going to start learning more about the process.   
  1. New developments and lower cost – This will allow for better data collection and faster deep learning.  Deep learning uses edge devices and servers.  As a result, there will be new servers, software and optimized systems that will speed up the learning process for the computer at that specific company.  Cost savings will occur using new startups who are starting to market in this area.  Companies need more and more storage for their data.  Alternate sensors and newer methods of gathering and using data are becoming more commonplace.  Higher demand results in lower prices.
  1. Developer tools for AI and deep learning keep on changing – TensorFlow is the most popular.  Caffe, PyTorch, and BigDL are becoming more popular as the demand grows.  New tools will help simplify the architecture and hyperparameter tuning, training, and disbursement.  Progress is expected in simulators that will speed up AI development.  Reinforcement learning libraries will improve also.  AI applications will be able to process multimodal inputs.  Most of all, there will be tools available for those who aren’t data scientists or engineers. 
  1. Use cases for automating – As more companies enter the realm of AI they will discover that automation is best used here or at least semi-automation.  This will include speech and natural language, robotics, and use cases in the health and medical fields.  AI will also be used in more creative fashion, like for music, art, and images.   
  1. Privacy, ethics and responsible AI – Transparency is in high demand now and will be a focus for a long time.  This is along with what is fair and what is explainable.   Responsible AI is key here.  Who wants to hear and see “fake” news all the time?   

Leave a Reply