Machine Learning, the future is here…

Being a science fiction freak, in the past when I heard AI (Artificial Intelligence), I thought about Skynet and robots offing humans to take over the world. Back then I would ask myself, do we want machines to make decisions for us and run our everyday lives? Especially ones that can learn and think more like us. I knew that computers would become more advanced, but I thought of AI as being a thing of science fiction and something that would not be around in my lifetime.

We don’t have killing robots trying to exterminate us, but AI is becoming more of a reality and is being used more often in our ordinary lives. More than we may realize. Our smartphones can unlock with face recognition. Smart speakers in homes can recognize voice commands, operate lights, thermostats, and door locks. Social media sites can recognize faces from pictures in postings and notify others of the posting. Business websites can learn users’ interests to target desired consumers based on what sites they have visited. The possibilities are endless.

Companies now realize the importance of AI in making quick decisions, to satisfy the endless demand to process data and to respond to that data quickly to get an edge on competitors. Because of the growing need for businesses to use AI in their systems, many Machine Learning tools have sprung up to enable them to implement it.

What is Machine Learning?
So, what is Machine Learning? Machine Learning is simply the science of programming computers to learn to act and make decisions, like humans do, by receiving data in real life scenarios. AI relies on data fed into the system to be used by advanced algorithms to make complex decisions. The more data the AI has, the better the chance that the AI will give an accurate response. Good AI will improve over time as it receives more data and learns from its own mistakes. Therefore, over time, the AI will become more refined.

What are the Benefits of Machine Learning?
As mentioned before, companies realize the importance of AI in making more accurate and timely decisions. With large amounts of data, AI can assist businesses by processing the data and then giving facts and suggestions for the company to use in making their final decisions.

Machine Learning can also be used to improve user experiences. Websites and social media can respond to user browsing based on their activity on the web in real time and use that input to make educated guesses about a user’s needs. Advertisements on company websites can then be customized to each user in response to his/her desires or needs.

Risks of Machine Learning
The benefits of machine learning are apparent, but there are also risks involved in Machine Learning. While Hollywood would have you believe that self-aware machines may someday take over the world, there are more realistic downsides to this technology. AI can only be as good as its data. In other words, if you feed it garbage, it will respond with garbage. Any flaws in the data will be reflected in the AI response. Sometimes, the data in a system even though accurate, could still give results that are undesirable and can negatively impact a company’s image.

An example of this is Amazon’s recruiting tool that showed a bias against women. In this case, Amazon was using ten years of resume data to create a tool that helped them to evaluate candidates for hire. Because the resumes came from mostly men the AI system thought that men were the most desirable candidates, thereby excluding many qualified women from the selection.

Machine Learning Tools Available
There are many machine learning tools available. New tools are being introduced and older ones are being upgraded constantly. I will provide some popular ones, but keep in mind that you may want to do some research on your own to see if there are more suitable tools for your particular needs. Most scripts in Machine Learning Tools are built in Python, but scripts can be built in other languages such as C++, Matlab or Java. If you want to investigate further on what languages can be used go to the following. https://www.purelogics.net/blog/best-programming-languages-for-machine-learning/

 

  • Azure Machine Learning Service
    Microsoft has just made some changes to its Machine Learning Tool. As of January 9th, 2019, Microsoft has deprecated three tools, (Machine Learning Workbench, Azure Machine Learning Experimentation and Model Management) and combined them to create a more centralized tool, Azure Machine Learning Service Workspace. In this one tool, you can experiment, train, and deploy machine learning models (mathematical representations of real-world processes). Microsoft has two different tiers in which they charge for their Machine Learning Tool, Free and Standard. There is no charge for the Free tier, but there are limitations to the service. The Standard tier charges a fee per seat per month and charges a fee per studio experimentation hour. You can see a complete list of charges by going to the following website. https://azure.microsoft.com/en-us/pricing/details/machine-learning-studio/.

 

  • Amazon SageMaker
    Amazon SageMaker uses a Jupyter notebook instance to experiment and analyze data and it contains common algorithms that are optimized to run very large data in distributed environments. After models have been trained, they can be deployed to produce predictions in one of two ways. A persistent endpoint can be set up to get one prediction at a time using Amazon SageMaker hosting services or predictions can be retrieved for an entire dataset using Amazon SageMaker batch transform. Training and hosting are billed by minutes of usage, with no minimum fees and no upfront commitments

 

  • Google Cloud Machine Learning Engine
    The Google Cloud Machine Learning Engine can train models by using the Google Cloud Platform resources. Google Cloud platform services can manage these models. The trained models are hosted on the Google Cloud Machine Learning Engine where they can receive prediction requests. Pricing is per hour and varies depending on the tier. Pricing is available by going to the following: https://cloud.google.com/ml-engine/docs/pricing.

 

Others
There are many other Machine Learning Tools, including open source tools such as Scikit-Learn, Shogun, Mlpack, and Accord.Net.

The world of Machine Learning is continuously changing, and new tools are being developed all the time. As this technology advances, companies will be able to use Machine Learning to not only predict customer desires and needs but also to be able to direct their advertising towards their target markets. The possibilities in the future are endless, and therefore, it is essential for information technology professionals to keep informed about the changes that occur in this growing area.

References Used on this Blog
https://www.reuters.com/article/us-amazon-com-jobs-automation-insight/amazon-scraps-secret-ai-recruiting-tool-that-showed-bias-against-women-idUSKCN1MK08Ghttps://docs.microsoft.com/en-us/azure/machine-learning/service/concept-azure-machine-learning-architecture#workspace

https://docs.microsoft.com/en-us/azure/machine-learning/service/overview-what-happened-to-workbench

https://docs.aws.amazon.com/sagemaker/latest/dg/how-it-works-deployment.html

https://docs.aws.amazon.com/sagemaker/latest/dg/whatis.html

https://cloud.google.com/ml-engine/docs/

Share this:

Leave a Reply

Your email address will not be published. Required fields are marked *