Machine-Learning-ApplicationsWhen someone mentions machine learning, what do you think of? Does your mind run off to dark and futuristic cinematic landscapes in which humanity’s robotic servants have rebelled and now wage war against their creators? If so, then you couldn’t be further from reality. Today’s most advanced artificial intelligence programs tend to do quite a bit more to help the human race, than they do to try and stamp it out. Take the various machine learning apps that are being developed, for example. These applications are capable of ingesting huge amounts of data and distilling it down into actionable facts. How does this help us? Well, let’s take a moment and look at a few examples:

1. Predicting launch success

Big data experts, hack/reduce, have been working in conjunction with dunnhumby to develop a system by which they can predict the success or failure of newly-launched products. This takes into account information such as the type of product being released and the number of stores selling the product. If successful, the resultant application could revolutionize business, and save the American economy millions (or even billions) of dollars in wasted marketing effort annually.

2. Battling credit-card fraud

Online credit-card fraud costs retailers approximately 3.5 billion dollars in chargebacks annually. As such, you can bet that there is a push to develop technologies that can better identify suspicious credit-card purchases and flag them for closer inspection. Sift Science, a San Francisco startup company, is attempting to do just that. Their application will be cloud-based, and will be available for use by other businesses to help them monitor their transaction in real time.

3. Identifying and assisting with potential health problems

When it comes to serious health-related issues, early detection is the key to survival and fast recovery. The problem is that in order for most illnesses to be properly identified before they become a major issue, a patient must be willing to submit to regular checkups and doctor visits—and given the current state of American health care, many simply will not. However, certain learning applications, such as the one currently in development by Professors Nathaniel Osgood and Kevin Stanley of the University of Saskatchewan, can help by collecting and analyzing important behavioral data, and using it to predict health risks in users. Another learning app called Just Shake It is targeted toward those who experience seizures or are at high risk of other health problems, such as strokes. This app allows users who are taken ill to contact emergency services by simply shaking their phone. The application uses machine learning to differentiate between the general tossing and josseling experienced by most smartphones, and the purposeful shakes of someone in need of medical help.

4. Improving personal identification and security

In order to really keep your personal information safe on a mobile device, you need to make sure that only those who are authorized can access it. This has lead to the use of password and PIN locking features, and even biometric fingerprint scanners. Still, these security measures all have their own flaws. In order to come up with something a bit more fool-proof, Seal Mobile ID is attempting to use the accelerometer found in most smartphones as a personal identifier. By learning the specific way that a user holds his or her phone, the application will be able to correctly identify and allow access without the need for PINS or Prints.

5. Protecting wildlife

Humanity continues to have an increasingly negative impact on various wild-animal species. Often times, this has less to do with any maliciousness on our part, and instead is a result of carelessness or other unintentional factors. Marine Explore and Cornell University have teamed up to create a learning application that will use whale audio recordings to help prevent collisions between marine mammals and transport ships. Others are hoping to use machine learning to be able to identify bird species sets from audio recordings, so as to get a better understanding of behavior and population. This will help conservationists determine if environmental changes are having a negative impact on local animal species.

So don’t judge all forms of artificial intelligence based on the actions of a few fictional, homicidal examples, because machine learning, and other advances like ultra-fast flash array storage, has the potential to do a lot of good. Let’s just make sure that we don’t put any of these new applications in charge of our nuclear weapons arsenal—after all, there’s no point in taking chances.