AI/ML: Amazing Insight / Magic Lessons
We have worked in transformative technology: Web, Mobile, Blockchain and AR/VR. In this article we will cover another major one, our journey in Artificial Intelligence (AI) / Machine Learning (ML).
Having worked in ML @ biotech for several years have see the power of ML first hand. Constantly amazed at what ML can teach us about our lab experiments. Amazing Insights (AI) / Magic Lessons (ML) … 😁
BioData ML Scientist (2019 — Present) - Cleaned, processed and analyzed biological data samples using machine learning.
- Data: Google Colab (Python) with Pandas & Numpy.
- Model: Optimize ML pipeline with TensorFlow & Scikit-learn
- Understanding: SHAP, LIME & ELI5
- Visualization: Ploty, Altair, Seaborn & Matplotlib
- Deployment: Plotly Dash in Docker image on AWS with 99.8% accuracy
Now it touches everything / everywhere we do from science fiction, design, mobile development to hard science.
Let’s explore our ❤️ of AI/ML starting from fiction (soft) to fact (hard).
AI/ML in SciFi
“Imagination is more important than knowledge.” — Albert Einstein
SXSW
Artificial Intelligence: From Fiction to Fact — SXSW Submission —
A tour of Artificial Intelligence(AI) in popular literature as compared to the very exciting recent advances in AI and Machine Learning happening right now.
AI has been a central player in science fiction & literature, all the way back to 1907 with the appearance of Tik-Tok in “Ozma of Oz”. Since then, robots and AI have suffused science fiction books, comics, and movies. These characters drove scientists and engineers to achieve the things they read in literature.
We will discuss the history of AI representations, from early robot appearances in Fritz Lang’s “Metropolis” and Karel Čapek’s “R.U.R.” to the appearance of room-sized AIs during the Golden Age, to the conversational AIs from the Silver Age such as Majel Roddenberry’s ship’s computer on the Enterprise, AI is everywhere.
Takeaways:
- Connecting popular representations in fiction to reality so the audience can see the contrast by showing cutting edge research from around the world.
- Demonstrate how to build your own AIs using free, open-source tools (TensorFlow, Keras, and Google’s Colab), and help inform your own fantastic work!
- Help creators find truth in their representations of modern AI, and let fans discover both the reality and the tools that make that reality possible.
SV Comic Con
Sci-Fi Comes Alive on Mobile — SV Comic Con Session —
Why is science fiction so important to science? Could we say science fiction writers are paving the way for future scientist?
To answer this we start by looking at the link between imagination and innovation, product design, technology, and core science culminating with a case study of how comic books are used to push innovation.
Digging deeper we review three major themes in Sci-Fi:
- Medical Devices
- Augmented Reality
- Artificial Intelligence.
We explore how they went from science fiction to advanced technology that used to only exist on supercomputers and now to running in everyone’s pockets because of the ubiquitous distribution of powerful smartphones.
We start with which companies are building “Medical Tricorders” and the applications which are running on mobile phone saving lives around the world.
Next, we look at what is AR and why it is considered the next big thing in tech by exploring the history and some mind-bending demos. After building a solid understanding of what and why we start diving into the how. We will explore early AR development and how it led to the current way AR apps are built with AR Core (Google Android) and AR Kit (Apple iOS).
In the last section, we look at what is Artificial Intelligence (AI), Machine Learning (ML) and Deep Learning(DL). We explore early ML development tools/frameworks and how they led to the current ML development environment MK Kit (Google Android) and Core ML (Apple iOS).
This entertaining talk starts by exploring real-world zombies and ends by examining the possible futures as more advanced technology makes its way to mobile devices.
The hope is that after attending these sections the audience will be well informed about the technology and have an appreciation of how ML, AR, and medical mobile apps are developed.
Movie / Films / TV
— Consulting AI/ML for Hollywood —
Wolff and Ash will take writers/directors/producers on a tour of artificial intelligence in popular literature and compare those to the very exciting recent advances in AI and machine learning happening right now.
Artificial intelligence has been a central player in science fiction and fantasy literature, all the way back to 1907 with the appearance of Tik-Tok in “Ozma of Oz”. Since then, robots and AI have suffused science fiction books, comics, and movies. These characters drove scientists and engineers to achieve the things they read in literature.
We will discuss the history of AI representations, from early robot appearances in Fritz Lang’s “Metropolis” and Karel Čapek’s “R.U.R.” to the appearance of room-sized AIs during the Golden Age, to the conversational AIs from the Silver Age such as Majel Roddenberry’s ship’s computer on the Enterprise, AI is everywhere. In the 80s and 90s, imaginations turn to scarier ideas with HAL and Terminator. In the most recent decade, an explosion of fruitful AI research has inspired friendly, helpful AI like Janet from “The Good Place”.
Each of these popular representations in fiction will be connected to the current state of art in ML so the audience can see the contrast between fiction and reality. Along the way, we’ll show mind-bending new AI/ML demos from the cutting edge of research around the world, including images, language generation, translation, medicine, question answering, robotics, and lots more.
We will also demonstrate the basics of building your own AIs using free, open-source tools (TensorFlow & Scikit-Learn using Google’s Colab), and help inform your own fantastic work portraying convincing and accurate AI representations in your own work.
::Consultants::
Wolff is an engineer on the Google Brain Team working on tools for machine learning. Before Google, he worked as a video game developer for 12 years. He has a PhD in artificial intelligence.
Ash works at a biotech using machine learning to analyze analytical data. He is a frequent speaker on topics of AI, AR and advanced mobile development. You can follow his talks on Twitter @Biocodes.
All our explanations will be in plain English for non-technical folks, but for experts we can scale up our discussion when asked.
AI/ML in Science (Courses)
School of Fashion
Making Machine Learning Fashionable — Fashion Course —
Tech class for the school of fashion (SF & NY)
Technology is the New Black ~ Ash
Course covers tech applied to the fashion industry.
- Artificial Intelligence / Machine Learning
- Blockchain / Smart Contracts
- 3D Printing
- Wearable Tech
Syllabus [work in progress with sections being added / updated constantly]
AI/ML in Science (Sessions)
Speaking around the country on AI/ML
Google Developers Group (SF & BK)
TensorFlow (ML) & ARCore (AR) — San Francisco & Berkeley —
AI with Pixel 4’s “Neural Core” Edge TPU & TFLite — GDG SF & BK
Wait … What ??? Google used an Android phone to build a self-driving car! Yes, in this section we will review how Google used a Pixel phone to build a “smartphone car” that can drive completely autonomously by using its camera/sensors to detect and understand signals from the world around it (sense lanes, avoid collisions, and read road traffic signs).
https://blog.tensorflow.org/2020/07/pixelopolis-self-driving-car-demo-tensorflow-lite.html
We will review the core concepts that were instrumental in building this amazing demo.
Video: ML section starts @ 25:00
ML Slides:
AI Dev World
Swifty TensorFlow — AI Dev World
Opening Conference Talk
In this talk, we look at “Why Swift for TensorFlow?”. The choice was guided by the goals of the project, which imposed specific technical requirements which we cover in this talk. A cornerstone of the design is an algorithm that is called Graph Program Extraction, which allows you to write in an eager execution-style programming model while retaining all of the benefits of graphs. The design also includes support for advanced automatic differentiation built directly into Swift.
We review the needed prerequisite concepts so that this talk is accessible to all technical levels.
Also presented at 360iDev
360iDev
Why Swift for TensorFlow — 360iDev Denver
In this talk, we look at the deliberation process that helped explain “Why Swift for TensorFlow?”.
The choice was guided by the goals of the project, which imposed specific technical requirements which we cover in this talk. We defined goals around the properties that are important to maintain and improve in the system: Expressiveness, Performance Predictability, Fast Iteration Time, Debuggability and Introspection, Flexible Deployment, Fast Deployment, Best-of-class Automatic Differentiation (AD), Embrace TensorFlow Graph Ecosystem.
A cornerstone of the design is an algorithm that is called Graph Program Extraction, which allows you to write in an eager execution-style programming model while retaining all of the benefits of graphs. The design also includes support for advanced automatic differentiation built directly into Swift. Following the theoretical section, we will do a Swift model training walkthrough.
By the end of the talk, the audience will have a good overview of what is Machine Learning, how TensorFlow works and why Swift for TensorFlow is the best way to train a model.
This talk is targeted at Swift programmers with no or little machine learning experience.
~~~
Credits:
Special Thanks to the Google TensorFlow Team!
