Should I use Machine Learning? – SC5

https://sc5.io/posts/use-machine-learning/?utm_content=59645558&utm_medium=social&utm_source=facebook

Did you know that without a single hour invested in development, you can approximate the usefulness of machine learning techniques in your product or digital service by answering two simple questions. Read and learn!

4 Comments

  1. Tomi Engdahl says:

    Y Combinator takes machine intelligence startups to school and learns a thing or two
    https://techcrunch.com/2017/08/27/y-combinator-takes-machine-intelligence-startups-to-school-and-learns-a-thing-or-two/?ncid=rss&utm_source=tcfbpage&utm_medium=feed&utm_campaign=Feed%3A+Techcrunch+%28TechCrunch%29&utm_content=FaceBook&sr_share=facebook

    The reality is that algorithmic advances become outdated on a nearly weekly basis in the world of AI. Things tend to go open source faster than they can even reach full deployment. This means that a startup initially using off the shelf AI tools might actually have a speed advantage to collect critical data over its competitors. This ability to forge gold from iron combined with meaningful domain expertise is the difference maker between successful and unsuccessful machine intelligence startups.

    big challenge is helping customers who purchase services from machine intelligence startups understand what it means to rely on a stochastic product. Outcomes aren’t always predictable and often they’re not even explainable.

    This is where I think the distinction between concrete and soft problems comes in handy. Concrete problems tend to be easily automatible. They are typically quantitative and highly repetitive in nature. Humans are very good at them but they are labor intensive. Try to think of standard classification problems like grouping images or extracting numbers from a document.

    Meanwhile, soft problems tend to be things that humans are not particularly good at.

    I wouldn’t trust an AI to look at my photo library and use the knowledge within it to write a letter to my mom.

    it’s important to remember the cost of making a mistake.

    Reply
  2. Tomi Engdahl says:

    The age of AI surveillance is here
    https://qz.com/1060606/the-age-of-ai-surveillance-is-here/

    For years we’ve been recorded in public on security cameras, police bodycams, livestreams, other people’s social media posts, and on and on.

    The time and effort it would take for someone to trawl through months of security footage to find a specific person, or search the internet on the off-chance they’ll find you is just unrealistic. But not for robots.

    Reply
  3. Tomi Engdahl says:

    Neural networks: Today, classifying flowers… tomorrow, Skynet maybe
    Here’s one I made earlier – now, over to you
    https://www.theregister.co.uk/2017/11/27/inside_neutral_nets/

    Reply
  4. Tomi Engdahl says:

    Three Things to Consider Before Incorporating Machine Learning into Your Security Efforts
    http://www.securityweek.com/three-things-consider-incorporating-machine-learning-your-security-efforts

    We have been hearing a lot of buzz about artificial intelligence (AI) for years, but more recently, the discussion within the cybersecurity industry has centered around machine learning (ML), an approach to AI that focuses on using algorithms to sift through data, learn from it and inform action based on the analytics, such as automatically preventing an unknown threat.

    When you unpack the history of AI/ML, you quickly realize the science behind it has been in development since the 1950s, or earlier, with Alan Turing’s seminal paper posing the simple question in 1951, “Can machines think?” But, if the methodology has been around for decades, the natural question is, why now?

    In order to maximize the effectiveness of ML in your security efforts, it is helpful to first understand what you need to do before adopting it. I would recommend that security practitioners focus on the following criteria when considering adding a capability that includes machine learning:

    1) Collect high-quality data – Having access to massive store of high quality data is the basis for training a machine learning system. When you adopt a product that includes ML, you will want to augment the things you have done in the past, like signature collection and automated malware analysis, so you can combine those things with the machine’s capability to determine new, malicious content. In addition to looking at bad data, you also need to have a large collection of good data, so that when it comes time to train the machine, it can accurately distinguish between what is dangerous and what is benign.

    2) Establish consistency in your security – Ultimately, you will need to ensure ML algorithms can run at multiple levels including network traffic, user behavior and endpoint. For example, if today you are only looking at anomalous behavior in your network traffic but not on your endpoint or in your user behavior, you won’t be able to accurately correlate and determine whether something is truly malicious so you can make the most sound decisions.

    3) Ask the right questions for vendors – Many companies claim their solutions incorporate ML, but oftentimes capabilities are overstated. The questions you ask these vendors should focus on how accurate, fast and efficient their systems are. Where does the analyzed data come from, and how often is it collected? How quickly can the solution make a decision that leads to an action? Developing and asking a comprehensive list of questions like these will help you select the system best suited for your company’s needs.

    Reply

Leave a Comment

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

*

*