How Machine Learning could help the Disabled


According to Tom Simonite(2017) :”Software that can understand images, sounds, and language is being used to help people with disabilities such as deafness and autism in new ways.”

Google’s Machine learning technology is offering ways that effective internet closed caption can be obtained to help those who are hearing impaired through speech to text recognition software. We all know that Youtube’s CC are not efficient, but recently (on the 23rd of March)they have released algorithms that now indicate applause and laughter. Hopefully, the new update has improved the captions, since the past system has been rather hilarious in it’s inefficiency as many youtubers have pointed out.

`Much research is going into similar projects by different companies. Facebook launched last year an image recognition to text description algorithm. IBM on the other hand is trying to better explain idioms of everyday use that are complicated to understand to those with learning disabilities. Research for new accessibility tools has also started to take   off in Europe.

Anything that makes the lives of those who are impaired easier is good in my eyes. Thoughts?

Machine Learning Won’t Take Over The Google Algorithm



According to Barry Scwartz(2017) from Search Engine Roundtable, Google’s Gary Illyes said that machine learning is impractical to use fully for Google’s ranking algorithms due to the fact that debugging would be a hassle. It does not mean that they are not using machine learning code in their algorithms, just that the majority will not be focused on it.

Schwartz states: “Debugging a machine learning process is too hard because machines do the work themselves. If someone hand codes an algorithm, it is way easier to debug issues with that.”

It is curious to me that one of the thing that is holding AI back is the debugging process of the code. Something that for me seems so trivial is actually a giant barrier for a machine to do on it’s own. I guess no Hal3000 anytime soon from Google. Thoughts?

Machine Learning Reddits

Machine Learning is a hard topic to understand. It’s concepts are advance and many that want to dwell in it find themselves lost. I decided to find a few subreddits that will support all that want to dive into this topic.

First: LearnProgramming Subreddit

This subreddit introduces all students to all type of questions and suggestions on how to learn programming. I suggest heading over to the FAQ section, read the rules and then study the different threads to make sure you do not ask anything that has already been asked.

Second: LearnMachineLearning

This subreddit introduces all students to all type of questions and suggestions on how to learn Machine Learning. It discusses general topics and you can also engage in the “best” ways to do things like: Pattern Matches. It is very educational and informative for new ways to do things are being discussed daily. Read the rules and then study the different threads to make sure you do not ask anything that has already been asked.

Third: MachineLearning

This subreddit discusses news about Machine Learning, it’s newest development, were it is being applied, and who is currently developing new programs, methods, etc…I sugest you look up the discussion on: “A Super Harsh Guide to Machine Learning”.

Fourth: MLClass

A Reddit study group for the free online version of the Stanford class “Machine Learning”, taught by Andrew Ng, previously shared in this blog. The purpose of this reddit is to help each other understand the course materials.

Fifth: DeepLearning

Deep learning (also known as deep structured learning, hierarchical learning or deep machine learning) is a class of machine learning algorithms. A Reddit with resources for understanding and implementing “deep learning” (unsupervised feature learning).

Sixth:  MachineLearning Jobs

This subreddit is basically a list of universities and companies that are currently looking for people to work in Machine Learning.


Machine learning will be a game-changer for esports

Read Article Here: Article

Video games and machine learning are merging. According to this article: “Game studios are applying machine learning to first address long-standing challenges in game development and design, like accelerating the process of building out art and levels and balancing the backend infrastructure.”

Specifically gaming is turning more personalized to adapt to each gamer specifically. It is ” being used to create progressional curves that tailor themselves directly to each player according to their individual behavioral data.”

How exciting is that? Games will slowly become slightly different for each player. Imagine big world games like the Fallout franchise changing more drastically the story according to your gaming style. For example, I love wandering, exploring and collecting items, scavenging if you will. That isn’t an easy task in Fallout 4, which loves for you to stick with story missions. Instead of the missions for the story being so straight line, with only a few options that differ it slightly, I would get scavenging missions that actually help in the development of the story and not just be side quests. A girl can dream…

What do you all think?


Machine-learning startups

Startups are all the rage now in business. Young entrepreneurs all around the world are putting their faith in these new small companies that are appearing in the dozens all around. But what about startups in machine-learning software? Are they a good investment? MEDHA AGARWAL,  from REDPOINT VENTURES believes so.

Big corporate giants like Google, Facebook and Amazon are using machine learning more and more these days, so there definitely is a market for it. From all the points that Agarwal mentions, I agree completely with him in the need of machine learning to predict certain future outputs. Today we generate so much data in 30min that a machine learning software  could easily have a high probability prediction of any type. I believe a startup geared towards prediction software that applies machine learning would be successful, especially in finance or insurance companies.


What do you all think?

Read The article: Opportunities for machine-learning startups: An investor perspective


Udemi online course on Machine Learning


The topics in this course come from an analysis of real requirements in data scientist job listings from the biggest tech employers. We’ll cover the machine learning and data mining techniques real employers are looking for, including:

  • Regression analysis
  • K-Means Clustering
  • Principal Component Analysis
  • Train/Test and cross validation
  • Bayesian Methods
  • Decision Trees and Random Forests
  • Multivariate Regression
  • Multi-Level Models
  • Support Vector Machines
  • Reinforcement Learning
  • Collaborative Filtering
  • K-Nearest Neighbor
  • Bias/Variance Tradeoff
  • Ensemble Learning
  • Term Frequency / Inverse Document Frequency
  • Experimental Design and A/B Tests

I found an online free Stanford course of Machine Learning


Here is a course taught by Andrew Ng of Stanford.

This course provides a broad introduction to machine learning, datamining, and statistical pattern recognition.

Topics include:

(i) Supervised learning (parametric/non-parametric algorithms, support vector machines, kernels, neural networks).

(ii) Unsupervised learning (clustering, dimensionality reduction, recommender systems, deep learning).

(iii) Best practices in machine learning (bias/variance theory; innovation process in machine learning and AI).

The course will also draw from numerous case studies and applications, so that you’ll also learn how to apply learning algorithms to building smart robots (perception, control), text understanding (web search, anti-spam), computer vision, medical informatics, audio, database mining, and other areas.