Let's start off the group by working through Fast.ai's 7 week course, Practical Deep Learning ForCoders, Part 1

The describe the course as:

Welcome to fast.ai's 7 week course, Practical Deep Learning For Coders, Part 1, taught by Jeremy Howard (Kaggle's #1 competitor 2 years running, and founder of Enlitic). Learn how to build state of the art models without needing graduate-level math—but also without dumbing anything down. Oh and one other thing... it's totally free!

We'll meet, do introductions and discuss Lesson 1: Image recognition. Please work through what you can from the lesson and come share your findings.

Deep learning (DL) has become an important topic in the AI and machine learning world. New advances in algorithms, hardware and software have made it a critical part of self driving cars, language translation, image understanding, medical analysis, and many other fields. Though DL systems are commonly written in Python (wrapping C/C++ libraries) we want to explore options available to Clojure developers.

We'll have a 2 part meeting with myself (Julio) and JR leading extended lightning-talk style presentations.

First, Julio will guide us through a discussion on:

• the basics of deep learning and the common tools and libraries • Java and Clojure options (DeepLearning4J and Cortex) and when they may be applicable.

In the second part JR will discuss his experiences while taking a deep learning for self driving cars course.

JR Says: Neural networks are just plain spooky. In just the past few years, they've gotten so good at classifying images (is this a cat or a dog? is this tissue cancerous or non-cancerous?) that they're often better at their task than humans are. In this talk, you'll see how to use a neural network - the same one used by comma.ai, a real-life self-driving car company - to drive a simulated car better than JR can.

website

Dr. Dejing Dou leads the Advanced Integration and Mining (AIM) Lab at the University of Oregon.

In this talk, we focus on the use of formal (Semantic Web) ontologies in two data mining and deep learning tasks, using health datasets as the data domain. We will show how Semantic association mining allowed discovery of indirect (hidden) associations within the datasets (in this case, among diseases and drugs within electronic health records). We will also show how Ontology-based deep learning was used to predict human physical activities within a health social network.

For more information and to register: https://damapdx.wordpress.com/2017/04/20/may-2017-semantic-data-mining-deep-learning-in-health-datasets/

Schedule

8:30 – 9:00 am – Sign In
9:00 – 10:15 am – Presentation
10:15 – 10:30 am – Break, Chapter Announcements
10:30 – 11:30 am – Presentation continued

Cost

Costs cover continental breakfast and speaker travel costs
- Free for Members (and employees of corporate members)
- $15 for Non-Members
- $5 for Students with valid student ID
See our corporate members at the link above

About DAMA Portland

The Portland Metro Chapter of the Data Administration Management Association has been serving the Portland data community since 1984. We are a not-for-profit, vendor independent, professional association dedicated to advancing the concepts and practices of enterprise information and data resource management.

Our primary purpose is to promote the understanding, development and practice of managing data, information and knowledge resources as key enterprise assets.

The DAMA Portland Chapter is dedicated to delivering thought provoking data-centric presentations that will make you more successful in your job. Please join me, the rest of the DAMA board, and other fellow lovers of data at this regular chapter meeting in downtown Portland. For more information and to register go to: http://www.damapdx.org/

Upcoming Presentations
June (15-Jun) Master Data Management
July ( 20-Jul) Data Science: Predictive Analytics
Sept (21-Sep) Cartographic license and building maps that work
Oct (TBD) DAMA Day (Our annual, full day of data training)

Thanks for your support of local Data Education, we look forward to seeing you there!

Ok, it's been a little over a month since the last meeting and I hope you are ready to get back into it.

Keep in mind this is a 'peer mentoring' style group. It does not matter what 'level' you are at. Please come with questions and projects and a willingness to discuss and help each other.

Lets start with the first video from part 2 of the Fast.AI course (on style transfer) and the first chapter of the Deep Learning Book (links below).

We're at a new location that should be quieter and have more parking. There is no food there but you can bring your own as long as you clean up afterwards. We'll see how it turns out.

Looking forward to it.

Julio

http://forums.fast.ai/t/part-2-early-release-videos-now-available/2621

http://www.deeplearningbook.org

This week we'll do Lesson 1 and Lesson 9. Pick which everyone you are comfortable with (or both) and work through the material and then come out and share your experiences and help others. For extra credit there are lots of variations you can implement as well as re-implement the code in pure TensorFlow and/or PyTorch.

Do you want to start using Kaggle? Why not learn as a group? I'll share the basics and an example how to start the competition "Titanic - Machine Learning from Disaster".

The second half of our session, we'll experiment with our open idea session. The winner will be awarded Innovator of the week. The winner has the highest potential to be added to our next session.

Here is the link to my transcript and video links. Enjoy! https://portlanddatascience.wixsite.com/home/single-post/2017/06/19/Kaggle-Session

This week we'll do Lesson 2 and Lesson 10. Pick which everyone you are comfortable with (or both) and work through the material and then come out and share your experiences and help others. For extra credit there are lots of variations you can implement as well as re-implement the code in pure TensorFlow and/or PyTorch.

This week we'll do Lesson 3 and Lesson 11. Pick which everyone you are comfortable with (or both) and work through the material and then come out and share your experiences and help others. For extra credit there are lots of variations you can implement as well as re-implement the code in pure TensorFlow and/or PyTorch.

We had many volunteers to talk about Python. We'll share the basics, Logistic Regression, and how to solve the Kaggle Titanic competition.

We'll have a few lightning talks then an open session for new topics.

You are welcome to join us and share your knowledge!

This week we'll do Lesson 4 and Lesson 12. Pick which everyone you are comfortable with (or both) and work through the material and then come out and share your experiences and help others. For extra credit there are lots of variations you can implement as well as re-implement the code in pure TensorFlow and/or PyTorch.

All aboard the Titanic. Your task, if you choose to accept it, pick a character on the Titanic. Then, you must build a machine algorithm to predict if you survived.

Now that you are afraid of dying, what's the best way to predict your survivorship?

Python Pandas and Numpy. If you want to learn how to build a machine learning algorithm check us out next Saturday at 1 pm in the Hillsboro Brookwood Library. We be using the Kaggle dataset to build our code.

This is a joint meetup with the CocoaHeads (iPhone) meetup. https://www.meetup.com/PDX-iOS-CocoaHeads/events/241700704/

We are going to explore a machine learning framework, Core ML, that Apple announced for the next version of iOS.

We had several new college students join us last week. We had a new idea to help college students with open data. Bring your college problems and we can help brainstorm how to help.

Did you know the top skill people want to learn is a language? How about a programming language like Python? Python is very powerful in Machine Learning. Check out our group on Saturday's at 1 PM.

We're working together to solve the Kaggle Titanic competition. Find out how you can learn Python and Machine Learning very easily and quickly.

At the end of the session, you'll get an opportunity to brainstorm and present new topic ideas for the following week. Did you know you learn exponentially more when you prepare a topic vs attending as a learner? The winning idea will get the innovator of the week.

Join the fun!

Did you know the top skill people want to learn is a language? How about a programming language like Python? Python is very powerful in Machine Learning. Check out our group on Saturday's at 1 PM.

We're working together to solve the Kaggle Titanic competition. Find out how you can learn Python and Machine Learning very easily and quickly.

At the end of the session, you'll get an opportunity to brainstorm and present new topic ideas for the following week. Did you know you learn exponentially more when you prepare a topic vs attending as a learner? The winning idea will get the innovator of the week.

Join the fun!

We have had a wonderful turnout of college students. It would be great to help with your projects. Please bring stop by and check us out.

Did you know the top skill people want to learn is a language? How about a programming language like Python? Python is very powerful in Machine Learning. Check out our group on Saturday's at 1 PM.

We're working together to solve the Kaggle Titanic competition. Find out how you can learn Python and Machine Learning very easily and quickly.

At the end of the session, you'll get an opportunity to brainstorm and present new topic ideas for the following week. Did you know you learn exponentially more when you prepare a topic vs attending as a learner? The winning idea will get the innovator of the week.

Join the fun!

Data Scientists,

Do you want to learn Python the Machine Learning Language?

or

Did you know MongoDB is the most popular NoSQL database? How interesting would it be if you could see it on a Raspberry Pi?

or

How about the schema of Stack Exchange the extremely useful site for stack overflow? Here is a link to Matt's presentation.

We have three excellent presentations this week. At the end of the session, we'll brainstorm a new interesting topic for the following week presentation. The winning idea will get the innovator of the week.

Presenter Topic Time Everyone Introductions 1:00 - 1:10 PM Matt SQL Project 1:10 - 1:35 PM Michael Raspberry Pi & Mongo 1:35 - 2:00 PM Garrett Python 2:00 - 2:20 PM Everyone Brainstorm & Vote Topics 2:20 - 2:30 PM

Do you want to learn Python the Machine Learning Language?

Do you want to see a live demo of an Arduino robot?

How about the Stack Exchange Schema?

We're exploring all of these innovative areas all in one session. We'll share a ton of knowledge in a short amount of time. Don't miss it!

At the end of the session, you'll get an opportunity to brainstorm and present new topic ideas for the following week. Did you know you learn exponentially more when you prepare a topic vs attending as a learner? The winning idea will get the innovator of the week.

Here's a link to my previous Python session https://youtu.be/OX2fVsKbwXg

Here's a link to our topic agenda https://docs.google.com/spreadsheets/d/15cmTfCJdwttpYEONdY-uGxMbCo7Z8_sWCKznKknGZ4k/edit?usp=sharing

Join the fun!

Data Scientists,

Do you want to learn about an unmanned robotic spacecraft Cassini?

How about programming a game in Python?

What about the data scheme behind Stack Overflow?

We have three amazing topics with code to share with the group. At the end of the session, we'll brainstorm next topics and someone will be awarded innovator of the week

Here's a YouTube of the previous session on Pygame. https://youtu.be/tRCYaCWzRnU

Join the fun!

Do you want to learn more about API calls to Google and Better Doctor?

How about Juno-Jupiter?

At the end of the session, we'll brainstorm new ideas and the winner will be crowned innovator of the week!

Here's a YouTube of the session from the previous week. https://youtu.be/U5sdLYHeJgk

Join the fun!

Agenda Presenter Topic Time Everyone Introductions 1:00 - 1:10 PM Kimberely API calls - Google Maps, Better Doctor 1:10 - 1:25 PM Everyone API calls - Q&A 1:25 - 1:30 PM Kateyln Juno-Jupiter 1:30 - 1:45 PM Everyone Juno-Jupiter Q&A 1:45 - 1:50 PM Everyone Brainstorm & Vote Topics 1:50 - 2:00 PM

Lets get together to discuss all aspects of deep learning and any questions you may have on the Fast.AI or Andrew Ng's coursera course.

All levels of experience are welcome. We rarely have presentations, though we may someone lead a discussion. Bring your questions, projects and ideas, basic or advanced, and we'll try to help.

Data Scientists,

Do you want to learn more about Data Visualization, BI, and Data Lakes?

How about more on Python Pandas and Data Frames? I'll be giving some examples of housing pricing trends.

At the end of the session, we'll brainstorm a new interesting topic for the following week presentation. The winning idea will get the innovator of the week.

Here's a YouTube of the session from the previous week. https://youtu.be/U5sdLYHeJgk

Join the fun!

Agenda Presenter Topic Time Everyone Introductions 1:00 - 1:10 PM Justin Data Visualization,BI,Data Lakes 1:10 - 1:25 PM Everyone Q&A 1:25 - 1:30 PM Garrett Python 1:30 - 1:45 PM Everyone Q&A 1:45 - 1:50 PM Everyone Brainstorm & Vote Topics 1:50 - 2:00 PM

Lets get together to discuss all aspects of deep learning.

In particular, this week I'm interested in learning more about ensemble methods for deep learning. If you get a chance and are so inclined check out The Relative Performance of Ensemble Methods with Deep Convolutional Neural Networks for Image Classification

Also, bring any questions you may have on the Fast.AI or Andrew Ng's coursera course.

All levels of experience are welcome. We rarely have presentations, though we may someone leads a discussion. Bring your questions, projects and ideas, basic or advanced, and we'll try to help.

Don't forget the #deep-learning channel in the PDX Startups slack which you can join at https://pdx-startups-slack.herokuapp.com

Join Tom Hanley from Skymind for an interactive tour through DeepLearning4j (DL4J), an open source, distributed, deep learning library for Java. This presentation will focus on machine learning basics and showing how to setup a development environment so you can run some of the canonical neurel network applications like image classification and text analysis with DL4J.

To get the most out of this presentation, read through the brief https://deeplearning4j.org/quickstart tutorial so you can follow along with coding examples.

Lets get together to discuss all aspects of deep learning.

In particular, this week I'm interested in learning more about AutoML (learning to learn). If you get a chance and are so inclined check out Using Machine Learning to Explore Neural Network Architecture and Google Vizier: A Service for Black-Box Optimization.

Our own Sylvain Payot will lead a short talk titled: AlphaGo: How DeepMind put an end to 2,500 years of human mastery

The game of Go has long been viewed as the most challenging of classic games for artificial intelligence combining an enormous search space and a challenging evaluation of board positions and moves.

In this talk, we walk you through the strategy adopted by Google DeepMind to develop the program AlphaGo, building on state of the art reinforcement learning and deep learning techniques to defeat some of the world’s greatest Go champions in 2016.

And finally, bring any questions you may have on the Fast.AI or Andrew Ng's coursera course.

Dear Learners,

Do you enjoy learning and sharing knowledge to help people solve problems? This week we're going to share about patents.

We'll have a thirty-minute talk and at the end of our session, we'll brainstorm and vote for a new interesting topic for ideas on upcoming presentations. The winning idea will get the innovator of the week.

Join the fun!

Lets get together to discuss all aspects of deep learning and any questions you may have on the Fast.AI or Andrew Ng's coursera course.

Also our own Matt Boyd will lead a discussion on:

"What I've found that works and can still be improved"

Matt will do a brief overview of the general approach to setting up a deep learning Kaggle project, touch on 2 or 3 good practices to follow, and throw around some ideas of what to improve.

Mat will cover how to index the data and build train/validation sets, finding the right balance of augmentation, and strategies for keeping results for 50+ different models straight.

All levels of experience are welcome. We rarely have presentations, though we may someone lead a discussion. Bring your questions, projects and ideas, basic or advanced, and we'll try to help.

Lets get together to discuss all aspects of deep learning and any questions you may have on the Fast.AI or Andrew Ng's coursera course.

We'll also focus on Hinton's capsules and see what we can make of them. Here are some useful links:

• The Arxiv paper

• A great youtube overview

• A Medium Piece

Can't wait to hear your experiences with capsules.

All levels of experience are welcome. We rarely have presentations, though we may someone lead a discussion. Bring your questions, projects and ideas, basic or advanced, and we'll try to help.

• What we'll do
We'll be watching and working through Artificial Intelligence/Deep Learning classes as a group! There are excellent free course lectures and materials online, and from January 6th 2018 to April[masked] we'll go through all the contents.

In order to cater to a diverse audience, there will be 3 structured sessions every Saturday – you can attend all, some or none, it’s totally up to you! If you don’t want to attend some of the sessions, throughout the day there will be open hacking on creating open-source code implementations of the top research paper pre-prints that week. Or use that time to catch-up on lectures and readings (sessions 2 and 3 have many hardcore readings by the way!) while discussing with peers.

Session 1: Fast.ai i Part 1 (v2) Lesson 2
Session 2: Stat385 Lecture 3 Readings – Overview of Deep Learning
Session 3a: UCL/Deep Mind Reinforcement Learning Lecture 2 – Intro to Reinforcement Learning

Note that Sessions 1 & 2 will begin with the 2nd lecture of their corresponding courses.

The Agenda
Session 1
10:00 AM - 12:00 PM (Beginner-Intermediate)
Practical Deep Learning- Fast.ai Part 1 (v2)
http://forums.fast.ai/t/welcome-to-part-1-v2/5787

Break
12:00 PM - 1:00 PM Lunch
Occasional brown bag lunch talk from an expert :)

Session 2
1:00 PM - 3:00 PM (Intermediate-Advanced)
Deep Learning Theory- Stanford STAT385 course on Theories of Deep Learning
https://stats385.github.io/readings
https://stats385.github.io/lecture_slides
https://www.youtube.com/playlist?list=PLhWmdj1YUpdT-UwCLVRNX509hZrKqZ83V

Session 3 (Intermediate-Advanced)
3:00 PM - 6:00 PM
Reinforcement Learning- UCL/DeepMind Reinforcement Learning
https://www.youtube.com/playlist?list=PLacBNHqv7n9gp9cBMrA6oDbzz_8JqhSKo

This is the Portland, OR chapter of AI Saturdays https://nurture.ai/ai-saturdays

Do you want to learn about Machine Learning for IoT devices? Chris is planning to join us fSensiML http://www.sensiml.com/

Do you enjoy learning and sharing knowledge to help people solve problems? Check out this video (https://www.youtube.com/watch?v=u6XAPnuFjJc) to see what we're about.

We'll have a short presentation and at the end of our session, we'll brainstorm and vote for a new interesting topic for the following week presentation. The winning idea will get the innovator of the week.

Join the fun!

Lets get together to discuss all aspects of deep learning and any questions you may have on the Fast.AI or Andrew Ng's coursera course.

Fast.Ai's Deep Learning For Coders Part 1 Version 2 has been soft launched and the videos, notebooks and code are available. So now is a great time to watch the first few and work through the exercises and come and discuss them. http://forums.fast.ai/t/unofficial-release-of-part-1-v2/9285

All levels of experience are welcome. We rarely have presentations, though we may someone lead a discussion. Bring your questions, projects and ideas, basic or advanced, and we'll try to help.

Also, if you want more structure and more content check out our own Brice's "AI Saturdays" https://www.meetup.com/AI-Saturdays-PDX/

Don't forget the #deep-learning channel in the PDX Startups slack which you can join at https://pdx-startups-slack.herokuapp.com

Do you enjoy learning and sharing knowledge to help people solve problems? Check out this video (https://www.youtube.com/watch?v=u6XAPnuFjJc) to see what we're about.

We'll have a 30-minute demo and at the end of our session, we'll brainstorm and vote on new interesting topic ideas for upcoming presentations. The winning idea will get the innovator of the week.

Join the fun!

This week Zach will walk us through The YOLO https://pjreddie.com/darknet/yolo/ neural net, how it works and how it can be used in your apps.

Lets get together to discuss all aspects of deep learning and any questions you may have on the Fast.AI or Andrew Ng's coursera course.

Fast.Ai's Deep Learning For Coders Part 1 Version 2 has been soft launched and the videos, notebooks and code are available. So now is a great time to watch the videos and work through the exercises and come and discuss them. http://forums.fast.ai/t/unofficial-release-of-part-1-v2/9285

Inference, Deep Neural Nets, Hyper-Parameters, Sign Language and Visual Insights

Welcome to TensorFlow Northwest. It was exciting to have so many of you come out during February’s snow-mageddon. This month, we prepared a balanced agenda to have new contents, and to incrementally build on what you’ve learned. On the other hand, we built the labs in a way that will allow those who couldn’t beat the storm to catch up quickly.

Al Kari will begin with an introduction and a technology update, then he moves on to cover a key topic in TensorFlow: Saving your models so you can use them to classify new data. The discussion will cover best practices and techniques to use your trained models for inference and transfer learning.

Al is Manceps’ Co-founder and CEO (www.manceps.com) where he helps enterprise clients with their Digital Transformation and intelligent computing roadmaps.

Next, Andrew Ferlitsch will take you on a code-along adventure via Colaboratory- An Introduction to Constructing Neural Networks and Hyper-parameter Tuning.

You’ve kicked the tires with 5 digits hand signs last month. Now, we progressively build layers of a neural network for the larger dataset of the American Sign Language hand gestures including all the letters of the alphabet. We will cover a fully connected neural network, proceed to a deep neural network, and then add dropout layers to tackle overfitting. Then we wrap up with the fundamentals of hyper-parameter tuning. The lab requires a laptop with a modern browser and will be at a modest pace to provide sufficient opportunity to ask questions and share insights.

As Chief Data Scientist at Manceps, Andrew works on solving a diverse set of problems using Deep Learning and advanced analytics.

Finally, David Molina will demystify Neural Networks once and for all, visually! What does bias do in a Neural Network? Why are weights important? What happens when we select a high learning rate? How do we know how many inputs to select? How many hidden layers a model needs in order to predict accurately? How does a Neural Network work? These are questions we ask when we begin studying Machine Learning as we encounter our first Neural Network. David will explore visually what’s behind the code, and what happens inside a Neural Network when it runs.

David is an Industrial Engineer with five years corporate experience in building databases, managing and analyzing large data sets, and optimizing systems and processes. He is currently studying Machine Learning and consulting for companies in his native country of Colombia.

REMEMBER to bring your laptop

Pizza and drinks sponsored by Intel. Links to the labs will be posted shortly.

Hope you have a great time learning and networking.

This is a joint meetup with TensorFlow NW and will be held at Vacasa. Check that meetup page for more information: https://www.meetup.com/TensorFlow-Northwest/events/246625002/

What interests you the most?

If you didn't make it to the TFDevSummit then you're in luck: We will bring the summit to you! A whole day of updates will be summarized for you in this session, where you get to hear what is available right now and where the technology is going next.

We are super excited to hold this joint event with the Portland Deep Learning group. In this interactive session, @alkari and @juliobarros will cover major topics and facilitate an open discussion on the latest features available in the framework.

For a bonus, @bhlmn will introduce Convolutional Neural Networks use cases in meteorology; Weather nerdery is a common phenomenon, and armed with a fresh PhD in the field, Bryan will get you thinking about CNN's potentials in Meteorology, now and in the future.

Come prepared for a dialogue and bring your questions, ideas and suggestions. There will be a quiz!

Don't miss this event.

Please register here: https://www.meetup.com/Data-Driven-PDX/events/250626939

AI & Machine Learning are delivering benefits to more organizations every day. We will explore:

Where do I start? How deep learning can be applied Best practices for getting started Decisions you need to make Proof of concept options Cloud, on-premises or hybrid? Resources available to you, and next steps We'll have beer, of course, food, networking, plus a great talk followed by Q&A featuring an AI expert from Skymind. Happy Hour starts at 5. The talk will start at 6 sharp.

Sponsors: Skymind, Pure Storage, Kovarus, Heceta Group

Speaker Biography:

Chris Nicholson CEO & Co-founder, Skymind (https://skymind.ai/)

Chris is the CEO and co-founder of Skymind, an open-source artificial intelligence company backed by Y Combinator, Ron Conway's SV Angel and Ray Lane. Chris worked as a journalist for over a decade, covering business and tech for The New York Times and Bloomberg News. He attended Deep Springs College, holds a degree in economics, and lives in the Bay Area.

Skymind is the leading open-source AI company, creator of Deeplearning4j and the Skymind Intelligence Layer (SKIL), an enterprise distribution for machine learning. It is Red Hat for AI, building machine-learning solutions for the Fortune 2000. Skymind supports companies applying AI to business problems that range from fraud detection to healthcare solutions to smart robots.

Can one lift the style from one image and transfer it to another? We will discuss the first paper to use neural networks to do this:

A Neural Algorithm of Artistic Style Leon A. Gatys, Alexander S. Ecker, Matthias Bethge https://arxiv.org/abs/1508.06576

We will go through the algorithm. I will present some example artistic transfers and discuss when it seems to work well and when it doesn't. I'd also like to get ideas from people about other use cases and generalizations.

If people are interested, we can also discuss practical implementation issues that may be glossed over by the paper. One implementation is available in the official pytorch tutorials. While it is a good place to start, there are some errors to look out for:

https://pytorch.org/tutorials/advanced/neural_style_tutorial.html

There also seem to be many TensorFlow tutorials available.

Bio: Currently Bill Dirks is a full time new dad. On the side, he is making art using machine learning. Previously, he worked as a software engineer for such companies as Google, New Relic, and Amazon.

This will be a hands on Python and computer vision machine learning so bring a laptop. We will be using the Fruits360 dataset from Kaggle to demonstrate how to prepare images for a variety of computer vision methods. The Gap framework uses a OOP style interface which should be familiar to application developers. Andy will cover how to use the framework to preprocess images, store/retrieve, regularization and mini-batch generation.

We'll get together to discuss questions you may have about deep learning. So don't be shy. Let us know what you are working on, what you've learned and what you'd like help with.

I also propose we read and discuss "A Survey of the Usages of Deep Learning in Natural Language Processing" https://arxiv.org/abs/1807.10854 Note: There is no single person presenting this paper we'll all read it and discuss what we learned and what we didn't understand.

== Slack == There is a #deep-learning channel in the PDX Startups slack which you can join at https://pdx-startups-slack.herokuapp.com

== Topic == Jonathan Mackrory will lead the discussion.

Music generation is one of the applications for deep learning where the emphasis is more on creativity than strict classification or regression. We'll have a presentation giving a quick survey of methods for music generation using deep learning. We'll have a whirlwind tour through representations of the music, important neural network architectures, some of the implementations, and hear some of the output. If we're lucky, some of it might even sound like music. We'll then wrap up with some open discussion about the topic.

The presentation will closely follow and be based on "Deep Learning Techniques for Music Generation - A Survey" by Briot, Hadjeres and Pachet. The article is available here: https://arxiv.org/abs/1709.01620 Note that the article is 108 pages, so there is little expectation anyone will read it. For those that want to look at the article, the essential material is in Ch 4,5,7,10.

Details
Same place, still looking for speakers. If you have anything you would like to present, let me know!

If you need parking, there's a parking deck below Safeway.

Agenda:

6:00 p.m.: Food, beverage, and networking

6:40 p.m.: Welcome message by Karl Fezer

6:45 p.m: Speaker 1: "Transfer Learning, or How to Stand on the Shoulders of Giants" - James DiPadua
7:30 p.m: Speaker 2: "Quantum Mechanics for Modeling Composite Semantic Spaces" - Connor Favreau
8:15 p.m.- 8:30: Project Ideas. Pitch your Project Ideas to this meetup group

8:30: End

Speaker 1 Details:

Abstract:

There's been a mountain of research into Deep Neural Networks' practical applications in image, audio and text processing. But these deep networks are often built on large corpuses of data (such as ImageNet or Wikipedia).

But that may not apply directly to your domain. Gathering data specific to your problem space may not only be a lengthy process but an expensive one too. That makes the business win a hard sell.

Transfer Learning can dramatically eliminate many of those problems, quickly.

In 'How to stand on the shoulders of giants,' we'll discuss the research background into Transfer Learning and how to implement the process in either Keras or PyTorch. The goal is for listeners to feel comfortable with the concept and prepared to begin researching an application in their workspaces.

Bio:

James is a wanderer, tinkerer, and ponderer. Not one to be pinned down, he's more comfortable in the abstract than in the known. He embraces ambiguity with a bearhug. That's a trick of course. He bear hugs the ambiguity into little mathematical boxes and then says "Dance!" and, oh, how that ambiguity dances! James currently hangs his hat at Vacasa where he works as a Senior Data Scientist tackling a myriad of growth-objectives with engineering and machine learning.

Speaker 2 Details:

Five years ago, Word2Vec offered a leap forward for the average data scientist to perform efficient algorithms in Natural Language Processing. From a body of text, Word2Vec generates a semantic space, in which the trained word vectors are often highly associated with their meaning. The next leap, a semantic space for phrases and sentences, proves tougher both computationally and in faithfully representing a composite meaning over multiple words. Surprisingly, quantifying particle interactions a la quantum mechanics shares close mathematical similarity to quantifying the meaning of words, phrases, and sentences. In this talk, I will provide an overview of current techniques in modeling language past word vectors, as well as point out the quantum mechanical aspects of these techniques. Emphasis will be placed on the “Compositional Distributional Semantics” model for the task of identifying word ambiguity."

We're back to our discussion format so there will be no presentation this month. Instead be prepared to share your insights and experiences on 1) privacy and ML and 2) The Fast.AI library.

Privacy for ML is a big topic so we'll start by reviewing the PySyft project, paper and tutorials. PySyft is a Python library for secure, private Deep Learning. PySyft decouples private data from model training, using Multi-Party Computation (MPC) within PyTorch. Links to materials are at: https://github.com/OpenMined/PySyft/tree/master/examples/tutorials

Fast.ai is a company dedicated to making the power of deep learning accessible to all. They've done this with their live courses at Unisversity of San Francisco, the corresonding free online MOOC, many free talks and presentations, and a deep learning framework built on top of PyTorch. I've heard many good things about the framework so lets dig into it and discuss it at the meeting. More info at: https://docs.fast.ai

Start your new year off right with other deep learning enthusiasts. Bring your questions and insights related to DL.

We'll also discuss PyText the new NLP framework from Facebook built on PyTorch. This will be a discussion topic not a formal presentation so review the package if you can.

https://github.com/facebookresearch/PyText

This month we'll be discussing an interesting use of Self-Attention Generative Adversarial Networks ( https://arxiv.org/abs/1805.08318 ) with DeOldify ( https://github.com/jantic/DeOldify ).

Please note, this is a discussion group not a presentation so grab the code, deoldify some photos, read the paper and come with your questions, insights and experiences.

Articulated pose estimation is one of the fundamental challenges in computer vision. Progress in this area can immediately be applied to important vision tasks such as human tracking action recognition and video analysis. This talk will discuss papers and progress of computer vision and deep learning towards human pose estimation and its applications.

Some papers that we can discuss here: Insafutdinov, E., Pishchulin, L., Andres, B., Andriluka, M., and Schiele, B. Deepercut: A deeper, stronger, and faster multi-person pose estimation model. CoRR abs/1605.03170 (2016). Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields, Zhe Cao and Tomas Simon and Shih-En Wei and Yaser Sheikh CVPR 2017

I will lead the discussion and would encourage everyone else to think about applications and future research/ development ideas in this domain.

Bio:

Srujana Gattupalli is a Deep Learning Software Engineer at Intel Corporation. She received a PhD degree in Computer Science from the University of Texas at Arlington in 2018. Her research interests are focused on Machine Learning, Computer Vision, Human-Computer interaction and their applications for human body motion estimation and pose tracking in assistive technology. Her academic work experience includes a role as a research assistant at the Vision Learning Mining lab and teaching assistant for graduate courses. She has been a Graduate Intern at Intel Corporation in 2017, working towards research and development for autonomous driving and machine learning algorithms. In addition to this, she has worked as a Software Engineer at Cerner Corporation in 2014. Ms. Gattupalli is an active member of Upsilon Pi Epsilon (UPE) honor society in computing. She has published 7 peer reviewed papers, received 2 international awards and has served as a reviewer in many others. In her spare time, she enjoys painting, philately, reading books, travel and to seek outdoor adventures.

As you know this is more of group discussion meetup where we get together to discuss what we are learning, interested in, stumped by, curious about, etc...

This month I propose we talk about Bayesian Deep Learning and study the materials from Yarin Gal's presentation at MLSS Moscow 2019 http://bdl101.ml

There wont be a presentation. Instead I encourage you to read over the materials, try out some experiments and come share your findings and experiences. I have lots of questions about when/where it is useful and what kind of future it may have. I'm looking forward to hearing your questions and thoughts.

All levels of experience and backgrounds are welcome so don't be shy. We all learn from each other.

Also, if you have ideas for topics you'd like to see in future meetings please let me know. Thanks.