New year’s resolution: dive into deep learning!

Fully Connected – a series where Chris and Daniel keep you up to date with everything that’s happening in the AI community. If you’re anything like us, your New Year’s resolutions probably included an AI section, so this week we explore some of the learning resources available for artificial intelligence and deep learning. Where you go with it depends upon what you want to achieve, so we discuss academic versus industry career paths, and try to set you on the Practical AI path that will help you level up. Sponsors: Fastly https://www.fastly.com/?utm_source=changelog&utm_medium=podcast&utm_campaign=changelog-sponsorship – 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 https://www.fastly.com/?utm_source=changelog&utm_medium=podcast&utm_campaign=changelog-sponsorship. Rollbar https://rollbar.com/changelog – We move fast and fix things because of Rollbar. Resolve errors in minutes. Deploy with confidence. Learn more at rollbar.com/changelog https://rollbar.com/changelog. Linode https://linode.com/changelog – Our cloud server of choice. Deploy a fast, efficient, native SSD cloud server for only $5/month. Get 4 months free using the code changelog2018. Start your server - head to linode.com/changelog https://linode.com/changelogAlgolia https://www.algolia.com/?utm_source=changelog&utm_medium=podcast&utm_campaign=changelog-sponsorship – Our search partner. Algolia’s full suite search APIs enable teams to develop unique search and discovery experiences across all platforms and devices. We’re using Algolia to power our site search here at Changelog.com. Get started for free and learn more at algolia.com https://www.algolia.com/?utm_source=changelog&utm_medium=podcast&utm_campaign=changelog-sponsorship. Featuring: Chris Benson – Website https://chrisbenson.com, GitHub https://github.com/chrisbenson, LinkedIn https://www.linkedin.com/in/chrisbenson, X https://x.com/chrisbensonDaniel Whitenack – Website https://www.datadan.io/, GitHub https://github.com/dwhitena, X https://x.com/dwhitenaShow Notes: Courses Amazon Machine Learning with AWS https://aws.amazon.com/training/learning-paths/machine-learningGoogle Machine Learning Crash Course with TensorFlow APIs https://developers.google.com/machine-learning/crash-courseMicrosoft Professional Program for Artificial Intelligence https://academy.microsoft.com/en-us/professional-program/tracks/artificial-intelligenceCoursera / deeplearning.ai https://www.deeplearning.ai/deep-learning-specializationNVIDIA Deep Learning Institute https://www.nvidia.com/en-us/deep-learning-ai/educationGoogle, Amazon, Microsoft: How do their free machine-learning courses compare? https://www.techrepublic.com/article/google-amazon-microsoft-how-do-their-free-machine-learning-courses-compareStanford Machine Learning Videos on YouTube https://www.youtube.com/watch?v=PPLop4L2eGk&list=PLLssT5z_DsK-h9vYZkQkYNWcItqhlRJLNBooks “Deep Learning” by Ian Goodfellow, Yoshua Bengio, and Aaron Courville http://www.deeplearningbook.org“Natural Language Processing with PyTorch” by Delip Rao and Brian McMahan https://www.amazon.com/dp/1491978236“Machine Learning With Go” by Daniel Whitenack and Janani Selvaraj https://www.amazon.com/dp/1789619890(Daniel is too humble to put his own book in this list, so Chris inserted it above without Daniel’s knowledge - because it’s a damn fine book!) Hybrid resources Machine learning with TensorFlow crash course https://developers.google.com/machine-learning/crash-course Intro to deep learning with PyTorch https://in.udacity.com/course/deep-learning-pytorch--ud188 Spinning Up in Deep RL https://blog.openai.com/spinning-up-in-deep-rl Meetups PyData https://www.meetup.com/pro/pydata Atlanta Deep Learning Meetup http://atlantadeeplearning.org Upcoming Events: Register for upcoming webinars here https://practicalai.fm/webinars!

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Speaker 00

  Bait with for change log is provided by Fastly. Learn more at fastly.com. We move fast and fix things here at change law because of roll bar. Check them out at rollbar.com and we're hosted on Lino servers. Head to linode.com slash change log.

• 0:21Speaker 00

  Welcome to Practical AI, a weekly podcast about making artificial intelligence, practical, productive and accessible to everyone. This is where conversations around AI, machine learning and data science happen. Join the community and slack with us around various topics of the show at change.com slash community. Follow us on Twitter. We're at Practical AI FM. And now into the show.

• 0:46Daniel Whitenack?

  This is Daniel Whitenack, a data scientist with SIL International, and you've joined us for another fully connected episode of Practical AI. In these fully connected episodes, my co-host, Chris, who's a chief AI strategist at Lockheed Martin, and I keep you fully connected with everything that's happening in the AI community. We take some time to discuss the latest AI news and dig into some learning resources to help us all level up our machine learning game. So, welcome, Chris. Good to talk to you again in in the new year.

• 1:20Chris Benson?

  Happy New Year, man. Good to talk to you, too. We got some exciting stuff coming up this year.

• 1:24Daniel Whitenack?

  Yeah, for sure. And this is a very special episode of Fully Connected. Normally, we go through a bunch of the latest news and that sort of thing, but we've decided for this episode to kind of give you our AI New Year's resolutions for the year and kind of go through our thought process of how we go about learning new techniques, new methodologies, new theory related to AI and deep learning. So this should be pretty fun. Are you excited, Chris?

• 1:56Chris Benson?

  I'm really excited, and I think this is a long time coming. We've been we're both very always advocating new learning resources and stuff. And I think it'll be fun to kind of share, share what we each have and kind of how we go about the process because I don't know about you, but I get asked that all the time.

• 2:12Daniel Whitenack?

  Yeah, there's so many resources out there and really there's such a wide variety of resources in terms of the background that they expect people to have, what tooling they use, what languages they use. And so it can be really overwhelming for people trying to pinpoint the right way to learn new subjects and dive into new things as related to AI and, and machine learning. So maybe before we jump into our resolutions, our specific resolutions, you know, Chris and myself, maybe we can just talk a little bit about the thought process that we go through when we're thinking about how to select the right

• 2:52Daniel Whitenack?

  sorts of resources. Now, in my mind, one of the things that I'm thinking about when I'm looking at resources is for me, there's resources out there related to deep learning and AI and other things that are really focused on research. And then there's a bunch of things really related to maybe the application of AI. Do you see similar trend there, Chris?

• 3:14Chris Benson?

  Yeah, I do. And I have a strong bias to acknowledge. And that is, I'm very much interested in the implementation side. I'm definitely not an AI researcher in that sense. And so I'm the kind of person who likes to come along. I program. As we're moving into the neural computing world, I love seeing some of these new capabilities coming out from all these different organizations, from Google to Amazon to Microsoft, you name it, there's so many. I like to find something that suits me. And that's the very first thing I do is, is to find something that's captured my interest in figuring out where I want to go.

• 3:48Daniel Whitenack?

  It sounds like you would be an excellent host for some type of practical AI content creation.

• 3:55Chris Benson?

  There you go. I, we should start a podcast.

• 3:58Daniel Whitenack?

  Yeah. So, yeah, I mean, I see the same thing. I think the first thing that, you know, maybe you want to be thinking of when you're trying to find new learning resources as you go into this new year are really what you want. And there's really no right answer to this question. Chris and I maybe. Lean more towards the application and integration and use of deep learning and AI and machine learning, but there's certainly a lot of like, we need great AI researchers. And so, you know, if you're leaning towards wanting to go into research, we certainly need people like that. So I think you want to be asking some of these questions like,

• 4:38Daniel Whitenack?

  do you want to do kind of deep learning research, figure out new sorts of AI, new sorts of neural networks or techniques that haven't been applied before and maybe do that in an academic setting, working on something very narrow for longer periods of time? Or do you want to kind of be in an industry setting or in some organization that is actually applying techniques in a more rapid time scale?

• 5:04Chris Benson?

  That's a great point because even though, you know, I declared my bias on the implementation, very much an industry focus, I can't do what I want to do unless brilliant people in academia or these days in industrial settings doing research are creating these amazing new tools that we can then new architectures that we can go apply. So lest I go too far on the implementation side, I definitely want as many people to go into research as possible selfishly so that I can, I can play with the fruits of their labors.

• 5:32Daniel Whitenack?

  Yeah, it's definitely needed. And so, I mean, as we shift to our resolutions, I think those are going to be much more application based and practical utilitarian, maybe if you want to say that. But if you're kind of leaning more towards the research side of things, often I get asked as I'm doing workshops and talks and other things, I get asked if, oh, if I'm going into this field or if I want to advance in AI, should I go and get a PhD? And as we've already said on the show many times, you don't need a PhD to work in AI. But if you're wanting to do sort of AI research and that sort of thing, you might consider other education. So you shouldn't do a PhD because it's a requirement to work

• 6:14Daniel Whitenack?

  in AI because it's not. In my opinion, you should do a PhD because you like research and you want to do research and that's often how that world works. So that's my. My opinion about that kind of further education. I don't know if there's certain education that you feel is relevant here, Chris?

• 6:32Chris Benson?

  No, I think that's for research. I agree with you. I would definitely say let that education be driven by your passion to do research. And I think in a lot of cases, going to that PhD is a great path because it's actually doing just that. It's a bit, but you're not getting the PhD just to do it. I mean, I don't have a PhD. I know you do. I don't have a PhD and I love this field and I'm able to be productive in it and lots of other people can. But at the same time, if someone's going to go into academia, I think that's, that's the probably the right path for them to take. Is to jump into a passion driven PhD program where they can go do the thing they love to do.

• 7:09Daniel Whitenack?

  Yeah. And there's also like, if you want to do, end up doing AI research in industry, maybe not being a professor. Like you want to work at Google Brain or OpenAI or one of these places. There are fellowships and opportunities for you to get involved there. So if maybe that's part of your New Year's resolutions for 2019 is kind of advancing to a goal of working with one of these companies, you can look up the OpenAI Fellows Program, the Google AI Residency. There's also some interesting things from the Allen AI Institute. They have actually a predoctoral kind of internship program. So that might be something if you're kind of starting to go that way, maybe you can look into that as something to work on in 2019.

• 7:52Daniel Whitenack?

  But assuming that you don't maybe want to go that route into research as Chris and I have not pursued that route as much. But maybe you want to just learn more about deep learning, dive deeper into deep learning. I think that, you know, it's perfectly fine to consider some self-study options, some kind of semi-guided study options. Maybe there's bootcamps, there's courses, there's a bunch of hybrid materials that include code practicums and exercises and videos and text pieces. So there's a whole lot of options out there, but maybe to start out and kind of narrow us in on, on some of the ones that we're interested in for 2019. Maybe this is a good time when we can kind of.

• 8:35Daniel Whitenack?

  Share some of our New Year's resolutions as related to AI. Mine are, you know, I've applied a lot of kind of, I don't know if I can use the word traditional, but more traditional sort of machine learning techniques, maybe regressions and kind of decision trees, random forests, a lot of these techniques in the past on various use cases. But some of the, I think I need to dive a little bit deeper into the neural network methods. I know we've talked a lot about them on the show and I know a lot of what's out there, but I'd like to dive deeper into some of those methods. Specifically, I think my resolution is to figure out some of the things that are going on with NLP

• 9:18Daniel Whitenack?

  and deep learning. So I specifically want to kind of dive into that side of deep learning in this new year and learn a little bit more about how that works and how I can apply it, especially since I work with an organization that's primarily concerned with language. Do you have any kind of specific New Year's resolutions or things that you would like to kind of level up on this coming year, Chris?

• 9:39Chris Benson?

  Yeah, I think, and it's different, it's a bit different this year than say last year at this time, because you know, as, as we have talked about, so many new tools and frameworks have come about and the capability of how far you can make it in a certain area without having to be just purely an expert in that area, that, that, that keeps getting easier and easier to manage from an implementation standpoint. And so I have a keen interest in robotics and I've, I've done that, I've been in that world professionally. One of the things that I'm, I'm very passionate about, aside from strictly work related, is doing stuff with my, with my daughter. And so one of the things this coming year that I'm planning to do is, is start to take some of the lower hanging fruits that are available on the NLP side and the machine vision

• 10:24Chris Benson?

  side and put them in some simple robotics things that I can share with her. That has me very, very excited, you know, totally outside of work and seeing her, and it's funny, she's already able to do that. So really, if you had asked me a year ago, it was very work related. And at this point, it's almost kind of bringing it home to some degree and being able to share it in more of a day to day kind of sense.

• 10:47Daniel Whitenack?

  Awesome. Yeah, I think that I always love hearing about your passion to make sure that you're both learning, but also able to kind of contribute and integrate, you know, your, your family relationships into, into this sort of work and with your daughter, learning all of these things. I think it's so cool. So, you know, with my resolution and kind of how I went about thinking about the resources that I'm going to target for this next year, I thought I want to learn about natural language processing and I want to kind of dive deeper into the deep learning methods for as related to language. So what I did was basically Google search, you know, NLP and PyTorch and TensorFlow and deep learning,

• 11:31Daniel Whitenack?

  and searched the O'Reilly website and searched other websites and Amazon and all of these things. And kind of the shortlist of what I came down to, and there are a lot of things out there, but there's a new book coming out from O'Reilly titled, "NLP with PyTorch", which seems very relevant to me because I've had some experience with PyTorch in the past, and it's relatively easier for me to understand in some respects. And so this is coming out soon, release date, January of 2019. And I'm excited. It covers apparently a bunch of things from recurrent neural networks to other things like LSTMs and, and other things. So I'm excited

• 12:13Daniel Whitenack?

  to maybe use that as a jumping off point to learning some more NLP with PyTorch. There's also a course on Udacity called "Deep Learning with PyTorch". It's not specifically geared towards NLP, but I thought it would be maybe good as I'm learning about NLP from the book, maybe I could dive into, maybe questions will come up around how this works in PyTorch, or why they did this with PyTorch. And so I think this course, which is free and covers things like the intro to PyTorch, along with related things like recurrent neural networks and natural language classification, I thought that would be a good supplement. And then finally, there's a set of videos from Pearson that are about deep learning for

• 12:57Daniel Whitenack?

  Natural language processing. And the reason why I, in my thought process, came to these videos was I thought, well, I don't necessarily only want to be versed in PyTorch. I also at least personally have the goal of being able to work both with PyTorch and TensorFlow, depending on the situation or the company I'm working with or whatever it is. So that one is actually more TensorFlow based. So I'd like to kind of learn, you know, both aspects of NLP and how it might be implemented slightly different in, in both. So that's kind of the thought process that, that I went through for my New Year's resolution. Does that make any sense to you, Chris?

• 13:34Chris Benson?

  Totally makes sense. And, and I'm not surprised, given the passions that you've expressed over time, where you're going and what you're interested in. So that's a fantastic set of resources to dive into.

• 13:44Daniel Whitenack?

  Yeah. So I realized by confessing all of these things on our podcast that probably some people will keep me accountable on, on these things. So feel free to do that in our Slack team or, or on LinkedIn and see how my progress is going. Or maybe if you want to go through some of these resources as I do, let me know. And in Slack, you can join our Slack team at change log.com slash community. Let me know. And maybe we can form a study group.

• 14:12Chris Benson?

  So when I got into, into my interest in deep learning a few years ago, there was a couple of, the resources were a lot fewer at the time in terms of choice. There's so much choice right now. And so for me, I really got started with a couple of things, one of which was, it was not when I started, but it came out not too far down the road was the deep learning textbook, which is, it's by Ian Goodfellow and I'm, I'm going to butcher the names, but Yoshua, Ben Guo and Aaron Corfield. Sorry about that middle one there. I've never gotten the right pronunciation, but it's a great aspirational read as you're getting into it. It's hard. It's truly a textbook, but it's something I'm constantly going back to as I'm trying to understand

• 14:55Chris Benson?

  the underlying mathematics and how the algorithms fit together. And I'll go to other books to pick up specific things that might be easier by O'Reilly or Pact Publishing or others like that. It's some of the basic concepts I tend to Come back to here and see if my understanding is better. So I really think it's a good investment as a it's kind of a metric on how your learning is coming along. If you can read the deep learning textbook and it makes sense on your basics, then you probably have it.

• 15:23Daniel Whitenack?

  Great reference for anyone.

• 15:24Chris Benson?

  So it's definitely worth the money. I have it literally right in front of me as I'm as I'm saying this right now and it's always nearby. And actually, there's another great book that I'm aware of, and since I know that you would never promote it yourself, I'm going to say it because it is a fantastic book. Daniel authors a book which is machine learning with Go and it is a really, really good book. I say that not just because I'm Daniel's friend, but because he's a fantastic data scientist and he's just written an excellent thing. And as the not author of the two of us, I want to note that you may notice that it is with Go and that's actually how Daniel and I met was we're both Go programmers called Gophers

• 16:03Chris Benson?

  in the Go community and got to know each other there first. And so I was absolutely thrilled when he wrote this book because I actually think, as does he, that Go is a great language to be able to do data science and artificial intelligence in. So since you would never Yeah, since you would never promote it yourself, I'm wanting to point that out. So I recommend everybody go to PAC Pub or whatever distributor like Amazon or other and get Machine Learning with Go by Daniel Whitenack. And I think it's a great book. I think I got my start early on. My first online course was the original Coursera machine learning course that Andrew Ngie taught and I got my certification.

• 16:42Chris Benson?

  But I would say that that one's pretty much out of date at this point. There's a more recent Coursera one on deep learning specifically. But I think as I go into 2019 and really focused on implementation and I really am trying to do a merge my software engineering background with my AI interests so that they really become seamless one. And so

• 17:05Daniel Whitenack?

  Which was one of the trends that we saw for 2019 and our last fully connected, right?

• 17:11Chris Benson?

  Absolutely. And it's I don't want them to feel like almost two different parts of me technically. And so I'm We're going to dive into the, um, the machine learning with AWS that Amazon has out at this point. And on that one specifically because I use AWS personally for lots of things. Um, and have been using it since I think 2010. So a long time. It's my kind of go-to framework. And then I'm also interested in both PyTorch and TensorFlow. And I want to do the Google's machine learning crash course as well with the TensorFlow API. So that's of keen interest to me going forward. And then I tend to supplement things with, with YouTube all the time. There's so many great videos on deep learning on YouTube. And so I have a friend who's actually been on our show,

• 17:53Chris Benson?

  Chris Dabellis, who is always telling me about the latest YouTube video in deep learning. And, uh, and that's a great place to go.

• 17:59Daniel Whitenack?

  Awesome. Yeah. And I think in light of, uh, kind of explaining how we, uh, have got to these resources as good resources for us, uh, this year. We might note that, you know, maybe I hope that these would be good resources for, for everyone, but they might not be the best for everyone. I think things to keep in mind while you're looking for learning resources this year. As you dive deeper into neural networks or whatever it might be. I think there's really kind of two types of routes that you might want to go. If you're coming from like a software development background, then maybe the challenging part of deep learning and neural networks and these things for you might be the more mathy things and the theory of it.

• 18:44Daniel Whitenack?

  Whereas if you're coming from kind of a scientist or academic background, maybe the challenge for you is not so much the math, but it's the kind of coding ability that you need to develop for some of these things. So if you're, if you're coming from a developer standpoint, maybe you want to pick out a resource that's really going to stretch you math-wise, right? And build up some of your skills on the math side of things. So maybe doing, uh, looking a little bit more into the theory of recurrent neural networks or, or something like that. Whereas if you're coming from the science or academic part, maybe that comes to you rather quickly. And maybe the thing that you want to focus on more is actually going through practical coding examples and making sure that the.

• 19:27Daniel Whitenack?

  Learning resources that you choose for the year or part of the year actually includes some practical exercises, some projects that you can work on through real code. For me, that latter one is generally what I look for, but I don't know. What's your filter as you're looking at resources, Chris?

• 19:45Chris Benson?

  That's a great point that you're making, and I probably should do more of that, actually, because I can actually relate that to a personal experience at a former employer. Our AI team really had kind of people that were software engineers who had converted over into this and people who were like data scientists purely straight out of school. And that's really what they did. And kind of that dichotomy that you just pointed out where, you know, developers may not have the math background that the data scientists had, but the data scientists sometimes were struggling with the programmatic and infrastructure issues. Each side going to what they're weaker in and kind of building up some skill on the other thing that you don't do is a really great thing to do.

• 20:28Chris Benson?

  Maybe I need to reconsider some of mine since I'm coming from the software engineering. I'm always trying to level up on my mathematics. Maybe I need to add that into my 2019 there. So, good point you made there.

• 20:39Daniel Whitenack?

  Yeah, and whichever side of that you fall on, there's definitely good intro resources on the math side. There's good intro resources on the coding side, like Python resources and other things. And so if you're struggling to find any of those, again, reach out in our community and we'd love to help you find some of those. But one of the things, I mean, learning resources, kind of books and courses can only, in my opinion, they can only get you so far. At some point, you need to actually apply what you're learning and work on kind of a side project or an interesting project of some type to actually try to apply some of what you learn. And that's really where I feel like a lot of things sink in and you gain a lot of knowledge.

• 21:24Daniel Whitenack?

  So in light of our resolutions, I thought it would be good to kind of give some examples of what a good kind of side project, good learning project would be like. I know Chris has some. I'll mention mine first. Since I'm kind of interested in the language aspect of things, there was some work recently by Facebook that did machine translation in an unsupervised way, which basically means that they didn't have a dictionary of this word corresponds to this word in each language or parallel language data, but they had monolingual data for each language. And then they did this unsupervised machine learning or machine translation technique.

• 22:04Daniel Whitenack?

  And I thought that was really cool, especially for people around the world that speak minority languages. And so the side project that I would love to work on as I'm going through these resources is related to that. I'd love to go, they released a GitHub repo that contains the code that they used in that. And I would love to maybe try that out on my own set of language data, or maybe even modify it to use some other word embeddings or something that we've talked about in past episodes, like the newly released embeddings called the BERT technique. And so I think my side project will be tinkering with some of that and see if I can get some machine translation going for these low resource scenarios.

• 22:49Daniel Whitenack?

  I think that would be a lot of fun. What do you think, Chris?

• 22:52Chris Benson?

  So a couple of things that in terms of getting hands on, I think in the past I've mentioned that I've been able to touch on the robotics world in terms of AI and robotics intersecting. And I think with my employer Lockheed Martin, there's some pretty amazing things that I'll be able to get involved in, and I'll talk about some of those down the road at the right time. But I think I'll address the personal side project things that I'm interested in. We're always talking about AI for good, I know. And as part of that, listeners probably don't know about me is the other thing that I do when I'm not talking technology is I'm really big into animal advocacy and animal rescue. It's actually how my wife and I met and our family is very, very active. That's what we do as a family is we go out and try to save critters.

• 23:36Chris Benson?

  So one of the things that I've wanted to do for a long time is to be able to use the machine vision technologies that are now available to be able to apply those to large scale. Maps like Google Maps and such and be able to detect dog fighting operations so that we can stop them. There are some fairly distinct characteristics and different types of operations and it doesn't necessarily have to be dog fighting, it can be other things. But with these new tools that nobody has had the opportunity to learn, I'm actually actively trying to recruit various law enforcement agencies in the United States have had a number of conversations and some other of the

• 24:13Chris Benson?

  animal advocacy organizations here in the US to form a coalition to pull together data sets and experts in the field and then apply the technology to it. So that's something I'm in the early stages, but I'm actively having those conversations. So I'm really excited to see what may be possible to end some, some suffering of, of animals that's quite horrible. So.

• 24:33Daniel Whitenack?

  Yeah, that's awesome.

• 24:34Chris Benson?

  That's a big part. And then the other one, which I've also mentioned previously is doing stuff with my daughter. I, I'm really excited about getting the stuff out of the office only environment and being able to bring it home and doing projects around the house. I have some of the, the normal things like I have a nest hello for the doorbell and, and some other things like that that are kind of IOT oriented. But I'm really interested in doing small scale projects that we can do as a family and they're actually useful. Cause this is not my, the AI stuff is not my wife's passion, but if I can produce something that's very useful for our family that my daughter's involved in, then that kind of brings it into the family environment. So that's, that's a big part of it is, is small drone and robotics related AI projects.

• 25:15Daniel Whitenack?

  Yeah. So as you're going into this year of, of 2019 and you're thinking, oh, I want to dive into this topic. Or I've found these resources that, uh, that I think are relevant for me. Think about some side projects that you're passionate about, like Chris said, maybe it's something that applies to your, your family situation. Maybe it's something that you're just passionate about in general, like animal advocacy or helping those with minority language speakers or whatever it is. So find something that you're passionate about and just try to get something working. I think that's a great way to learn these subjects. If you don't know where to start in terms of finding a good project to work on or.

• 25:57Daniel Whitenack?

  Or data related to a certain project that you're interested in. There's a couple of resources that I think are relevant and maybe you have one too, Chris. The ones that came to my mind are Driven Data and DataKind. Both of these organizations run either competitions or support projects that are utilizing AI for good in some sort of way. So I would recommend looking there. Maybe that's some inspiration for you to find some really good uses of AI to work on on the side. But also, as I mentioned, for kind of my target side project, maybe you've seen something in the news or on a previous Fully Connected episode, some kind of new advanced type of AI or maybe just a new result.

• 26:42Daniel Whitenack?

  Likely there's a GitHub repo that reproduces some of that. So maybe one way you could start diving into that is just by trying to get the code to work. Maybe subbing out the data that they use and trying out your own data. That might be a good starting point as well for you to work on the side. What about you, Chris, as you're kind of working on these robotics things and the AI for good applications? Do you have trouble with the data side of things or other aspects? What recommendations do you have for people?

• 27:11Chris Benson?

  Yeah, so myself, I never have trouble with kind of the inspiration for a new project, you know? Because it's really driven by what I want to do, you know, something that I'm passionate about. The thing that I'm always trying to do is find the right data sets that can contribute. And often it's not one data set. It might be a combination of different data sets that can, taken together, give me the use case that I'm trying to achieve. And so this last year, Google announced Google Data Search, which I think is just fantastic. Because you know, prior to that, if you hear about data sets, I kept bookmarks on different data sets around. But Google Data Search has revolutionized that by you can type in the keywords of what you're interested in in different ways and come up with the data sets that are publicly out there.

• 27:54Chris Benson?

  There are many, many thousands of them, far more than I could have ever bookmarked on my own. And so I think it's just, I think it's a must have tool that anyone in this field is going to be using on an ongoing basis. So Google Data Search is, it gets a big thumbs up for me.

• 28:08Daniel Whitenack?

  For sure. So kind of recapping what we talked about so far, you might have some resolutions like us for this new year. You know, use these filters that we've talked about, find some good resources for yourself, talk to us in the community. We'll try to guide you to that stuff. Find some good side projects to work on to apply what you're learning. And then I think the last thing that is really important in terms of learning is, is community. And I know, Chris, you definitely agree with this. You're involved in the Atlanta Deep Learning Meetup and, and things. Do you have some inspiration for, for listeners as far as why they should get involved in, in meetups or what benefits that those might, those might be in terms of learning?

• 28:52Chris Benson?

  Yeah, I do. And I would urge people, I'm the organizer, I created and organized the Atlanta Deep Learning Meetup. When I started it, which was at the very beginning of 2017, when I kind of started it saying, well, no one's, there's some general machine learning stuff, but I really want to have a deep learning conversation. And I don't know if anyone will ever show up, but I just kind of went for it. And it really, suddenly there was an outpouring of people who said, this is something I'm interested in. And they show. And we ended up, I think we have something like 2,200 people in the group now. Obviously, they don't all come, but yeah, but it's not uncommon for us to have 100 people show up for a particular monthly event. And we do it pretty much every month. And so what it's really done, the, the outcome of that isn't just about the presentations

• 29:35Chris Benson?

  at the meetup. It's, it's great when you have great speakers doing really interesting topics, but people often say, what about filming it or whatever? And they're almost missing what I think is probably the most important aspect. And that is getting in a room and talking to other people who share that passion and interest from other organizations or academia. There is so much value in those conversations and creating those relationships that no matter how good the presentation is, it's almost incidental to forming those relationships with different people. And on a larger scale, conferences are the same way. If you're going to go to a data or AI oriented conference, don't just go to the presentations, walk up and introduce yourself, be bold and have conversations with as many different people as you can, because some of those will really take you places you

• 30:20Chris Benson?

  weren't expecting that were wonderful. So engaging other human beings in this world of AI is about one of the best things you could possibly do for yourself.

• 30:28Daniel Whitenack?

  Yeah, and, and you know, we're talking about learning here. Everybody at those meetups, everybody at those conferences is learning. They might know a lot about a small number of subjects, but they don't know a lot about many subjects, and that's true for, for every person. So you're wanting to learn about some things, you know, it's perfectly fine to, to ask questions and engage in discussion because likely all of those people there are having, some of the people are having similar questions. You can get connected to people that already know about that thing, but maybe they don't know about some of the things that you know about, and so it is really useful. One of the things that I want to do this coming year, since I'm, will hopefully be

• 31:10Daniel Whitenack?

  diving a little bit more into this topic of, of NLP, is I've been to a lot of meetups, a lot of conferences related to kind of, uh, operationalizing AI and machine learning. That's mostly been my, my discussion at those conferences. So I've gone to a lot of, you know, infrastructure-related conferences and other things, along with machine learning conferences, but I've never actually, you know, participated in really a, uh, more kind of cutting-edge research sort of conference community like NeurIPS or EEM NLP or, or one of these. So maybe it would be cool this year to at least, um, attend or at least try to submit

• 31:51Daniel Whitenack?

  something based on my learnings with, with NLP to, to one of those conferences. I think that would be a part of the AI community that I haven't interacted with as much, but I think I could learn a lot from them, and so I would be excited to, to get involved there. You would do very, very well at that, Daniel. I think you would rock the house. Well, I, I appreciate that. I, it'll be interesting. I'm, I'm interested to, just excited about a lot of, that's going on in this community. I know you are as well. It's, it's exciting to be, to see things moving so fast and have so many opportunities to dive in to, to interesting topics. You know, I'm just excited about learning a bunch this year.

• 32:31Chris Benson?

  Yeah, I think we are fortunate to be able to work in what is surely the coolest field on the planet right now. It is so fast moving. There's constantly new things to learn. You never get established. You never get to a point where I finish learning. You just wait a week or two and there's, there's the next thing. So if you love to learn, if you love to constantly be moving, it's a great field to be in either professionally or as an amateur. Either way, I definitely encourage people with the interest to dive into it.

• 33:00Daniel Whitenack?

  Yeah, definitely. And you know, we have limited time on the podcast, so we're only able to share a few of the resources that are on our mind this coming year, but we've got a ton of learning resources that we know about and we're going to list out a whole bunch of those in the show notes of this episode. So make sure and check those out. Find something that, that you can work on this year and level up your skills. And if none of those things in the show notes, you know, makes sense, and even if they do make sense, again, join our community on Slack, join our community on, on LinkedIn and participate in the discussion around, you know, things you'd like to see on the show, but also maybe learning resources

• 33:40Daniel Whitenack?

  that you're looking for and people can help you, help you find them. So excited about 2019. Looking forward to, to learning more.

• 33:48Chris Benson?

  All right. Happy New Year, Daniel, and Happy New Year to everyone out there. We have a great year ahead.

• 33:53Daniel Whitenack?

  Yep. Bye bye.

• 33:57Speaker 00

  All right. Thank you for tuning into this episode of Practical AI. If you enjoy the show, do us a favor, go on iTunes and give us a rating, go in your podcast app and favorite it. If you are on Twitter or a social network, share a link with a friend, whatever you gotta do, share the show with a friend if you enjoyed it. And bandwidth for changelog is provided by Fastly. Learn more at fastly.com. And we catch our airs before users do here at changelog because of Rollbar. Check them out at rollbar.com/changelog. And we're hosted on Linode cloud servers. Head to linode.com/changelog. Check them out. Support this show. This episode is hosted by Daniel Whitenack and Chris Benson. Editing is done by Tim Smith. The music is by Breakmaster Cylinder. And you can find more shows just like this at changelog.com. When you go there, pop in your email address, get our weekly email

• 34:42Speaker 00

  keeping you up to date with the news and podcasts for developers in your inbox every single week. Thanks for tuning in. We'll see you next week.

• 34:55Speaker 03

  I'm Tim Smith and my show Away From Keyboard explores the human side of creative work. You'll hear stories sometimes deeply personal about the triumphs and struggles of doing what you love. I got really depressed last year and the reason it was so hard is because basically everything culminated at once. All these things I'd been avoiding, all these things I'd swept under the rug, they all came out at once. New episodes premiere every other Wednesday. Find the show at changelog.com/afk or wherever you listen to podcasts.