My goals are just to work on interesting and challenging problems in computer science. I'm in the text analytics field right now, which has lifetimes of interesting unsolved problems in it. I did some cool machine learning work in college I'd love to return to some day (Genetic Programming for building Traveling Salesman Problem heuristics, Image Edge Detection using Genetic Algorithms). Finally, I'm interested in Robotics, as I believe it's going to be the next huge industry.View the profile
About the talk
Today, most mainstream applications of AI are dealing with language. Chatbots, virtual digital assistants and automated transcription services are all powered by a collection of Natural Language Processing (NLP) technologies which enable users to mine text for meaning.
NLP includes sub-fields like Natural Language Understanding (NLU), which enables functions like translation and sentiment analysis, essentially turning text into structured data, and Natural Language Generation (NLG) - used to turn data back into human-readable text.
To find out more about working with NLP, AI Business sat down with Paul Barba, chief scientist at Lexalytics, a company that specializes in text analysis.
Barba is a machine learning expert who spent 12 years at the company, progressing from his position as a development intern to overseeing its research and development efforts. He was shortlisted for the AI Innovator of the Year award at the AI Summit New York in December.
Welcome to Addison's TV here at the anthem at New York 2019 be very pleased to be joined today by po Baba. He's a cheap blinds and electrolytic. How 00:04 are you? How are you enjoying a time he is so foggy 00:13 for a eye in a vase with the yeah. 00:20 I really appreciate the list was really meant a lot. I see a lot of great information going on every day, 00:30 and it's really nice to be recognized as a company for that. So, thanks. Tell me what kind of outfits inspired you to answer the tick world and a Ian 00:39
to pick him up. So early for a lot of us. There was some question that caught our attention of the kids, you know, some big question. Where do we go 00:48 when we die or why were dinosaurs so cool something that like inspires our philosophy and I remember looking up at a tree is real small kid and a warm 00:55 summer day and seeing all the leaves and being like, huh? Why am I seeing this to my cousins over there? And he seems something else like why am I me? 01:02 Why did my why was I born as him? And what would that mean? It is Billy the people I want having this experience. And so that we should have kicked 01:11
off interest in Consciousness and thoughts. And where do these arise from in between that and I really love programming a little bit older. So if 01:17 those two are just like some kind of brought me to this field and if you're really cool if you got to work in it cuz it's really interesting stuff. 01:23 You're kind of a i robot look like they what kind of project are you working on 01:28 currently. He's a little bit about the car. LP various acronyms overtime but we really try to use these tools 01:38
to solve business problems and always seen over the years is that it really comes down to these photos Monday and things are getting a really good 01:48 pipeline a good process having good tools and to build tools to automate and make it as easy as possible to have two Castle Project. So I feel really 01:56 lucky to work in a bit of a horizontal company where I told her everything it's working with fuel companies in question. Answering is working in 02:04 financial companies following regulations and everything. So we really focused on how do you solve these problems to the pragmatically? What is the 02:10
stuff you need to do to have a successful project? 02:20 It's always been 02:23 the same two things to see the people buying to the height and they're like, yes, it would be so easy. It's just going to solve all my problems going 02:33 to fill data scientist problems gone if you do everything and those projects off and fall over. The other thing you see if people have done just let 02:39 you know what I see the hype. It's icy over promise. I'm going to none of this real it's all smoke and mirrors and just had to turn away from the 02:46
field. I don't need to worry about it. My business would be affected and I think of the long-term that's even more dangerous, The hype is capital hot 02:52 sometimes gets a little ahead of itself but fundamentally that's is changing the world and you're seeing these real incremental steps that are 02:59 changing business. So one of the other criteria at 2 to get to whether or not enough 03:05 does the best solution the best best route to go and 03:09 lock his to say I kind of project strolling favor go 03:18
first. I didn't you have hit these problems before and really the biggest challenge is just 03:25 how many problems you can begin a run into the AI algorithms a really great at making use of any thing wrong with your data. Already a little 03:35 pattern that you didn't intends you you sampled incorrectly at some point and you'll get really promising results that will gen burst into flames in 03:45 production or you're interacting with an Enterprise you're fighting with humans in the state of Nigeria World with technology. We've always just kind 03:52
of put it in a box and then gone in and interact with its terms, but they are pulling in data sources to maybe a CRM system the customer has and then 04:00 some other part of the Enterprise has not changed their young vendors and so suddenly the date of when you're older than you've gone and everything's 04:08 on fire and how quickly can you find that how quickly can you retrain everything get it up again in the wrong and it's a critical they will be but if 04:14 you going to go to write tools, if you think about it ahead of time you can you know, you can if you can get past it you can have a strategy to have a 04:23
successful projects and End times I'm looking to the Future then what's going to be shaken up and surprise in terms of AI in the next 2 to 5 years. 04:30 What is going to be on everybody's lips? Like we going to be pause the hype now. This is going to be the next next kind of big Steak n Shake Up. 04:39 So maybe it's just my unbiased view but I'm we got at least three trains language models now for 04:46 Elmo And I think these advancements isn't the first year suddenly all the standard data. Do you know 04:56
but then it sort of this long tail of the next two to five years were kind of integrates the Enterprise will start using any more problems. 05:04 Understanding the nuances of Technology going to be a lot of cool new things. You can do with language other emails or social media or 05:13 regulations, you know, this text is such a fundamental part of who we are as humans and how we communicate at the foot of the defining factor of 05:22 humanity. So I think that's going to exciting area 05:30
Buy this talk
Buy this video
With ConferenceCast.tv, you get access to our library of the world's best conference talks.