Machine Learning By Alex Scarlatos
What is Machine Learning? Machine Learning is the process by which computers can be trained through observation, rather than being explicitly programmed. Basically, a program is given input and adjusts its behavior based on that input. While ML is an important part of Artificial Intelligence, they are not the same thing. AI is the ability for a machine to “think” on its own, and many research fields are encompassed by AI.
Example - A Personalized Music Station 1. The station picks a random song - “Baby” by Justin Bieber. 2. The user gives the song a dislike. 3. This is interpreted as -1 points for pop music and male singers. 4. The station then picks a random song, but sways the randomness so there is some less chance of getting a pop song or a male singer - it picks “Rain Is A Good Thing” by Luke Bryan. 5. The user gives the song a like. 6. The algorithm gives +1 points to country music and gives a point back to male singers. The station remembers user choices, and when a trend appears it will be accentuated. If the user keeps liking country songs and disliking other genres, the station will eventually only pick country songs.
Where we see ML today
https://techcrunch.com/2016/09/06/ultimate-guide-to-the-news-feed/
Microsoft’s ML Twitter Experiment In 2016, Microsoft released a chatbot named Tay onto Twitter. People could tweet at it and it was supposed to respond in relevant ways, and was meant to evolve through these interactions.
http://www.theverge.com/2016/3/24/11297050/tay-microsoft-chatbot-racist
So what is ML actually good for?
Medical Diagnoses Computer algorithms can analyze millions of medical cases and draw conclusions based on strong data patterns. When a new patient comes in, the program can compare their condition and medical history to previous cases and make a diagnosis. This is expected to give much more accurate predictions when determining a patient’s remaining time or chance to live.
Protein Folding Given previously found folds, a program can predict new folds by considering where others were successful. This can save researchers time and money. http://www.mdpi.com/1422-0067/17/12/2118/htm
Autonomous Cars The car is “trained” by drivers. It records what directions people turn in certain situations and will make its own decisions based on analysis of those choices. https://www.coursera.org/learn/machine-learning/lecture/zYS8T/autonomous-driving
So where does this lead us?
How can ML affect our culture? Machine Learning algorithms are capable of actually generating music. While this is unlikely to entirely replace human writers, a good portion of the music industry could theoretically become automated. “People have to get the idea out of their head that music comes from great individuals. It doesn’t, it comes from communities, it comes from societies. It develops over many years and computers become a part of societies.” - George Lewis, music professor https://www.washingtonpost.com/news/innovations/wp/2016/06/06/googles-computers- are-creating-songs-making-music-may-never-be-the-same/?utm_term=.4d8cdfda622a
Will robots actually take our jobs? By analyzing the ways that humans perform tasks, machines can gradually replace us at these tasks. This mostly applies to simple tasks that are repetitive and easily describable. http://www.bbc.com/news/technology-34066941
This might not all be bad news though Technology has been advancing rapidly since the Industrial Revolution, and people have always been afraid of being replaced. Technology improves quality of life for workers and consumers. We have always adapted to our technology, and while the future is unpredictable we shouldn’t expect this time to be very different. https://www.youtube.com/watch?v=UqESogRgrYw
What can we take away from this? 1. ML isn’t the answer to all of our problems... (Do we really want as susceptible as Microsoft’s Twitter bot making national security decisions?) 2. Many modern jobs will be replaced. But this is inevitable and could actually improve quality of life if society is adjusted properly.
Resources http://www.dataversity.net/ai-vs-machine-learning-vs-deep-learning/ http://www.theverge.com/2016/3/24/11297050/tay-microsoft-chatbot-racist https://www.statnews.com/2016/10/03/machine-learning-medicine-health/ http://www.mdpi.com/1422-0067/17/12/2118/htm https://www.coursera.org/learn/machine-learning/lecture/zYS8T/autonomous-driving https://www.washingtonpost.com/news/innovations/wp/2016/06/06/googles-computers-are-creating-songs-maki ng-music-may-never-be-the-same/?utm_term=.4d8cdfda622a http://www.bbc.com/news/technology-34066941 https://www.youtube.com/watch?v=UqESogRgrYw
Recommend
More recommend
