When data leakage turns into a flood of trouble
48 min

Rajiv Shah teaches Daniel and Chris about data leakage, and its major impact upon machine learning models. It’s the kind of topic that we don’t often think about, but which can ruin our results. Raj discusses how to use activation maps and image embedding to find leakage, so that leaking information in our test set does not find its way into our training set.

Discuss on Changelog News

Join Changelog++ to support our work, get closer to the metal, and make the ads disappear!


  • DigitalOcean – DigitalOcean’s developer cloud makes it simple to launch in the cloud and scale up as you grow. They have an intuitive control panel, predictable pricing, team accounts, worldwide availability with a 99.99% uptime SLA, and 24/7/365 world-class support to back that up. Get your $100 credit at do.co/changelog.
  • Changelog++ – You love our content and you want to take it to the next level by showing your support. We’ll take you closer to the metal with no ads, extended episodes, outtakes, bonus content, a deep discount in our merch store (soon), and more to come. Let’s do this!
  • Fastly – Our bandwidth partner. Fastly powers fast, secure, and scalable digital experiences. Move beyond your content delivery network to their powerful edge cloud platform. Learn more at fastly.com.


Notes and Links

Towards Data Science
Towards Data Science
The TDS team
60. Rob Miles - Why should I care about AI safety?
Progress in AI capabilities has consistently surprised just about everyone, including the very developers and engineers who build today’s most advanced AI systems. AI can now match or exceed human performance in everything from speech recognition to driving, and one question that’s increasingly on people’s minds is: when will AI systems be better than humans at AI research itself? The short answer, of course, is that no one knows for sure — but some have taken some educated guesses, including Nick Bostrom and Stuart Russell. One common hypothesis is that once an AI systems are better than a human at improving their own performance, we can expect at least some of them to do so. In the process, these self-improving systems would become an even more powerful system that they were previously—and therefore, even more capable of further self-improvement. With each additional self-improvement step, improvements in a system’s performance would compound. Where this all ultimately leads, no one really has a clue, but it’s safe to say that if there’s a good chance that we’re going to be creating systems that are capable of this kind of stunt, we ought to think hard about how we should be building them. This concern among many others has led to the development of the rich field of AI safety, and my guest for this episode, Robert Miles, has been involved in popularizing AI safety research for more than half a decade through two very successful YouTube channels, Robert Miles and Computerphile. He joined me on the podcast to discuss how he’s thinking about AI safety, what AI means for the course of human evolution, and what our biggest challenges will be in taming advanced AI.
45 min
The Artists of Data Science
The Artists of Data Science
Harpreet Sahota
Creativity is a Mindset | Nir Bashan
Nir Bashan is a world-renowned creativity expert. He has taught thousands of leaders and individuals across the globe how to harness the power of creativity to improve profitability, increase sales and ultimately create more meaning in their work. FIND NIR ONLINE Website: https://www.nirbashan.com/ LinkedIn: https://www.linkedin.com/in/nirbashan/ Twitter: https://twitter.com/Nir_Bashan QUOTES [00:05:42] "I wanted to be insanely creative, like some of my heroes and and I noticed that they had a technique, all of them a little bit different, but they all had some technique. And it turns out that that technique makes them no different than you or I or anybody listening to this podcast. And it makes them no different. It's all about coming up with that creativity and learning how to harness it and repeating it and pulling it out whenever you need." [00:08:32] "We live in a society in a day and an age where we can get data on just about anything...But what does it do to have that data? If we're not able to make that information work for us, it's meaningless. " [00:12:42] "We've gotten so comfortable in modernity that we're looking for problems that don't exist" [00:17:00] "A good company and a good person on a career path is armed with creativity to be able to solve problems as they occur in new and different ways. And that really is the most important thing that we can do today to become relevant and to stay relevant and to stay successful." [00:22:49] "I feel that quantification is just the mere act of assigning a number to something and that is just not good enough anymore. It's not." [00:25:51] "The numbers aren't really telling us all that we need to know. There's got to be a different way. There's got to be a new way to look at those things. And that way is creativity." [00:39:25] "Creativity is a component of our DNA and who we are. And in order for us to really be fully evolved human beings, I think we need to balance our analytical drive with the creative drive" [00:53:41] "Egos kill creativity, just like the self-doubt or perhaps even more. We develop egos when we think that we know what will happen and when we think that we have some advantage that other people don't have." SHOW NOTES [00:01:52] Guest introduction [00:02:53] Nir’s journey and path to now [00:05:31] Being creative isn’t just for artists and musicians [00:07:42] Why is it that just having logic alone is not going to be enough to propel us to the next level? [00:10:09] How can a non-creative person discover their creativity and put that to work? Harpreet Sahota:[00:14:29] How has creativity impacted business acumen? [00:18:14] What are some of the practices of the of yesteryear that we need to get rid of to really be successful in the business of tomorrow? [00:21:30] Talking about quantification, can you just kind of make that concept concrete for us? What is it that that you mean when you say quantification? [00:25:24] How you’ve been educated out of the creator’s mindset [00:27:16] The Trinity of Creativity. [00:29:15] Why is it crucial that we shut out the analytical part of our mind? [00:32:56] What creativity means to Harp Nir Bashan: [00:35:06] Why you shouldn’t be afraid to ask “stupid” questions [00:38:12] The importance of being open and receptive and willing to wander off the path that you've paved for yourself [00:39:17] Emotional intelligence and creativity [00:41:04] Positivity and creativity [00:43:58] What questions can we ask ourselves when we’re coming up to a problem to make we frame these problems in more positive ways? [00:46:04] What are the personality traits of creativity? How do we cultivate those traits for ourselves? [00:48:24] How do we catch ourselves when we're having negative internal dialogs so that we can shift ourselves to have more of the cultivated personality traits for creativity? [00:51:24] Psychological safety and creativity [00:54:33] How do we fight something (the ego) that comes naturally to us? [00:55:51] The Four P’s of growth [00:57:10] Why discipline is a pathway to creativity [00:58:18] How we can use a creative mindset to be more creative with our time management? [00:59:36] It's one hundred years in the future: What do you want to be remembered for? [01:00:50] The Random Round Special Guest: Nir Bashan.
1 hr 11 min
IT Visionaries
IT Visionaries
Robots IRL: The Future of Robots in the Workplace
What if instead of saying, “There’s an app for that,” the saying was, “There’s a bot for that?” And what if, instead of spending hours of your day performing repetitive tasks, a bot beneath the surface of your screen was doing the work for you? _“With RPA, you're able to make the lives of people at work much, much easier. It improves their own lives, and improves the end customer's life. It improves their business. It improves the value prop they provide. So that is the way I think about it. when you bring [RPA] together and say, I won't sell you A.I., or I won't sell you a cloud-native product. What I will sell you is an experience that allows you to do the things that you want to do in a way that makes sense to you and will be a delight for you to use and your end-user.”_ That’s Prince Kohli, the CTO of Automation Anywhere, a global enterprise RPA solution platform that brings robotic process automation to industries worldwide. On this episode of IT Visionaries, Prince discusses how RPA is changing the game when it comes to office efficiency. Plus, he details the future of RPA and whether we’ll see a future where everyone eventually will have a digital assistant. Main Takeaways * iRobot: Robotics Process Automation is not about physical robots completing tasks. Instead, RPA is automated software that builds off a set of rules in order to complete repetitive tasks, such as invoicing. The software works in the background on computers, allowing employees to be more productive and creative with their time. * Discovery and Auditing: There are two core functions the Automation Anywhere teams follow when they begin working with a client. The first is the discovery process, when they go in and observe what processes could be automated. The second is through auditing, when they gather insights and data in order to build a secure bot * There’s a Bot for That: In the future, most employees will have a digital assistant in some capacity, whether that is for filling out calendars or answering emails the technology is already there. It’s adoption that is the issue. --- IT Visionaries is brought to you by the Salesforce Platform - the #1 cloud platform for digital transformation of every experience. Build connected experiences, empower every employee, and deliver continuous innovation - with the customer at the center of everything you do. Learn more at salesforce.com/platform
46 min
Clear search
Close search
Google apps
Main menu