Internet of Things by Neil Wilkins

Author:Neil Wilkins
Language: eng
Format: epub
Tags: Internet of things, electronics software sensors actuators data IoT, Internet of things Reinforcement IoT ML DL ANL, analytics machine learning Artificial intelligence, artificial neural networks Recommender Systems, For business Risk Assessment Cyber cyberwar, Ebook ebooks book books paperback paperbacks
Publisher: Neil Wilkins
Published: 2020-06-16T00:00:00+00:00

Chapter 10 – Machine Learning

If a plant seed is placed in the dark and there’s even a hint of sunlight, the plant will grow, twist and contort itself as much as needed to reach the light. If a seed is placed in a dark maze and the plant needs to solve the maze to reach the light, it will do that as well[45]. We can arbitrarily scale the maze up, and the plant will keep struggling to find the exit, sending offshoots down separate paths for information on how to reach the light. How about placing a potted plant next to a window and rotating it away from the light to see what happens? The plant will slowly rotate itself back around, so the leaves absorb the most sunlight[46]. This happens imperceptibly and, with the exception of sunflowers, we scantily notice that plants can turn around and that they have a preference for the direction they’re facing.

How about a slime mold placed in a maze with a piece of food at the center? If the slime solves the maze, it gets a tasty treat, which it invariably does[47], and again we can scale the maze up, and it will always get solved in the most energy-efficient manner. Fungi, mice, birds, cats, dogs, elephants and chimpanzees – every living creature shows the same innate propensity for solving spatial challenges to reach food, and even humans plopped down in the middle of a shopping mall with a grocery list will eventually amble their way out the door with a loaded cart. All creatures except robots, thinking machine servants.

The pride and joy of human creation, the pinnacle of mechanical engineering, and yet, robots are as dumb as rocks and can’t do anything unless specifically instructed to do so through code, a set of machine-readable instructions. Any change in the environment invalidates previously written computer code; any conflict in the code leads to unpredictable behavior, which is what we call “bugs”. While living creatures have the genetic code to guide them through life’s challenges and mazes, robots and computers have nothing of the sort unless someone writes out a specific set of commands: if A, do B unless C. This means a robot has to have a specific code written out for every given maze, and the code needs updating whenever the maze changes or the robot is moved slightly from its starting position, or there’s any change in the environment whatsoever.

Machine learning is the brilliant idea that, since living creatures have the genetic code that holds instructions and computers have code too, perhaps creating such machines that can randomly mutate their programming can lead to something intelligent, the same way millions of years of evolution led to slimes and plants solving mazes in search for food. So far, there’s been just enough progress to whet the appetites of scientists working on the concept, but there’s no way to break through the conceptual barrier and create actual, independent intelligence. It’s tantalizingly close and yet it appears reaching actual artificial intelligence could be the undoing of us all.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.