In today's video, we use hands-on code demos in Python to find the slopes of curves with the Delta Method. While finding these slopes, we derive together — from first principles — the most important Differential Calculus formula.
This video is part of a thematic segment of videos on Differentiation. In the forthcoming videos, we’ll cover derivative notation and a series of useful rules for differentiation.
New videos are published every Monday and Thursday. The playlist for my "Calculus for ML" course is here.
More detail about my broader "ML Foundations" series and all of the associated open-source code is available in GitHub here.
How to Thrive as an Early-Career Data Scientist
Getting started in data science? Today's episode is for you! Sidney Arcidiacono is absolutely crushing her first year in the field; we discuss the options for getting started in the field and top tips for early-career success.
Trained as a phlebotomist (blood-sample collection), Sidney was inspired by the potential for machine learning to revolutionize healthcare, so she jumped feet first into a full-time computer science degree at Make School, specializing in the data science track. From no familiarity with code or models just a year ago, Sidney's immersion has paid off: She's now fluent in the modern data science software stack and landed a summer data science internship at GreenLight Biosciences, Inc., an RNA-molecule therapeutics firm (like the Pfizer/BioNTech/Moderna vaccines).
Sidney is terrifically sharp and engaging; I think you'll enjoy hearing from her as much as I did during filming.
Watch or listen here.
The Delta Method
In today's video, we use a Python code demo to develop a working understanding of the Delta Method, a centuries-old technique that enables us to determine the slope of a curve.
This video is the first from a thematic segment of videos on Differentiation. In Thursday's video, we'll build on what we covered today to derive — and deeply understand — the most common, most important equation in differential calculus.
New videos are published every Monday and Thursday. The playlist for my "Calculus for ML" course is here.
More detail about my broader "ML Foundations" series and all of the associated open-source code is available in GitHub here.
Peer-Driven Learning
"Peer-driven" learning — where you are formally taught by your coworkers — not only results in team members learning key new skills, but can have added benefits like team bonding, confidence, and innovation. Something to try!
Today's episode is directly inspired by a LinkedIn post by Laura Rodriguez. She tagged me in the post, citing a SuperDataScience episode on communication and relating it to her workplace at ForwardKeys. Thank you, Laura!
Derivatives and Differentiation — Segment 2 of Subject 3, "Limits & Derivative
Today marks the beginning of a new thematic segment of videos in my ML Foundations series. This segment builds on the Limits content already covered to clearly illustrate how Differentiation works and how we find Derivatives.
Through a combination of color-coded equations, paper-and-pencil exercises, and hands-on Python code demos, the videos in this segment instill a deep understanding of how differentiation allows us to find derivatives.
More specifically, the videos cover:
• The Delta Method
• The Differentiation Equation
• Differentiation Notation
• Rules that enable us to quickly calculate the derivatives of a wide range of functions, including those found throughout machine learning
New videos are published every Monday and Thursday. The playlist for my Calculus for ML course is here.
More detail about my broader ML Foundations series and all of the associated open-source code is available in GitHub here.
Learn my ML Foundations Curriculum LIVE on O'Reilly
From July through December, I'll be offering my entire Machine Learning Foundations curriculum as a series of 14 live, interactive workshops via O'Reilly online. It'll be a lot of fun — would love to see ya there!
Each training is 3 hours long with an extra 30 minutes afterward for answering your questions and working through exercises together. The curriculum is filled with hands-on code demos in Python, particularly the NumPy, TensorFlow, and PyTorch libraries.
There are four subject areas:
• Linear Algebra (3 classes)
• Calculus (4 classes)
• Probability and Statistics (4 classes)
• Computer Science (3 classes)
All 14 training dates are provided at jonkrohn.com/talks.
Sign up opens about two months prior to each class; the first three classes are open for registration now at oreilly.com.
The 20% of Analytics Driving 80% of ROI
Today’s episode is with freakin' David Langer, people!! (So obviously it's brilliant, witty, and full of laughs.) He fills us in on the most powerful 20% of analytics — the analytics that drive 80% of companies’ return on investment.
Publishing under his Dave on Data brand, Dave's YouTube channel is top-notch, with several videos that have over a million views (and the thumbnails are hilarious; check 'em out). He is an exceptionally accomplished data scientist and software engineer, including spending nearly a decade at Microsoft's Global HQ, where his titles included principal software architect, principal data scientist, and director of analytics.
Topics in the episode include:
Surprisingly powerful modeling approaches in spreadsheet tools like Excel
The SQL databases we'll need if the data sets we're working with are too big for spreadsheets
Why R programming is easy and should be our default language choice for moderate to advanced statistical analysis
How companies can maximize value from machine learning
Listen or watch here.
Exercises on Limits
Final YouTube video from my thematic segment on Limits out today! It's a handful of comprehension exercises. Starting Thursday, we'll begin releasing videos from a new Calculus segment, on derivatives and differentiation.
We release new videos from my "Calculus for Machine Learning" course on YouTube every Monday and Thursday. The playlist is here.
The Machine Learning House
In last week’s Five-Minute Friday, I discussed how, in the data science field, the learning never stops. But there’s one big counterpoint: The foundational subjects that underlie the field barely change at all, decade after decade.
These subjects — linear algebra, calculus, probability, statistics, data structures, and algorithms — build a strong foundation for your “Machine Learning House”. Today's Five-Minute Friday articulates my perspective that investing time in studying these foundational subjects will reap great dividends throughout your data science career.
You can listen or watch here.
Calculating Limits
Today's video introduces Limits, a key stepping stone toward understanding Differential Calculus. This one has lots of interactive Python code demos and paper-and-pencil exercises to ensure learning the subject is both engaging and fun.
We release new videos from my "Calculus for Machine Learning" course on YouTube every Monday and Thursday. The playlist is here.
Six-Hour "Calculus for Machine Learning" Tutorials
Descending a gradient of cost is what allows most machine learning algorithms to, well, learn. In this hour-long video, I derive cost gradients using partial derivatives and we use them to implement ML models in Python.
The free lesson is an excerpt from my new, six-hour Calculus for Machine Learning video tutorial, which is available via subscription to O'Reilly or via purchase from Pearson.
Machine Learning at NVIDIA
This week's guest is absolute rockstar Dr. Anima Anandkumar, who's both professor at prestigious Caltech and director of ML research at NVIDIA. The episode is exceptionally content-rich but also lots of fun; Anima was a joy to film with.
In the episode, Anima fills us in on:
The cutting-edge interdisciplinary research she carries out (applying highly optimized mathematical operations to allow state-of-the-art ML models to be executed on NVIDIA's state-of-the-art GPUs)
How this blending of leading software and leading hardware enables world-changing innovations across disparate fields, from healthcare to robotics
What it's like in the workweek of a researcher bridging the academic and industrial realms
The open-source data science tools and techniques that she most highly recommends
Listen or watch here.
Calculus Applications
New YouTube video out today! In this one, I provide specific examples of how calculus is applied in the real world, with an emphasis on applications to machine learning.
The YouTube playlist for my "Calculus for Machine Learning" course is here.
The Learning Never Stops (so Relax)
It's common to feel overwhelmed by the vast ocean of tools and techniques you could learn in the fast-moving field of data science. No need to stress though — it's all part of the job, so ease into it and have fun!
Listen to, or watch, today's Five-Minute Friday episode here.
Calculus of the Infinitesimals
New YouTube video up! In today's we use a hands-on code demo in Python to see how approaching a curve infinitely closely enables us to determine the slope of the curve. This is key to formally understanding differential calculus.
The YouTube playlist for my "Calculus for Machine Learning" course is here.
TensorFlow 2 versus PyTorch
Which of the two leading automatic-differentiation libraries — TensorFlow 2 or PyTorch — should you use for your deep learning models? My opinions, bolstered by recent usage data, are detailed in this talk that I gave at MLconf in November.
Thanks to Courtney Burton and Richard Rivera for inviting me to speak. It was an honor!
Abstract
This talk begins with a survey of the primary families of Deep Learning approaches: Convolutional Neural Networks, Recurrent Neural Networks, Generative Adversarial Networks, and Deep Reinforcement Learning. Via interactive demos, the meat of the talk will appraise the two leading Deep Learning libraries: TensorFlow and PyTorch. With respect to both model development and production deployment, the strengths and weaknesses of the two libraries will be covered — with a particular focus on TensorFlow 2 release that formally integrates the easy-to-use, high-level Keras API into the library.
99 Days to Your First Data Science Job
He's BAAAAACK! Kirill Eremenko is the GUEST on the SuperDataScience show for the first time. In this episode, he details his exceptional new learning pathway that enables folks to land their first data science job in 99 days.
We also cover:
• What Kirill's been up to; life philosophies he's honed
• 5 myths holding people back from starting a data science career
• 5 items you need to land a data science job
Kirill created the SuperDataScience podcast in 2016 and hosted the program (over 400 episodes!) until passing the torch to yours truly on January 1st.
Kirill also founded the SuperDataScience company and is the firm’s CEO today. SuperDataScience.com, the namesake of this podcast, is a comprehensive online education platform for data science and related data specializations. Through SuperDataScience.com and his Udemy courses, Kirill has taught well over a million students worldwide, launching countless data science careers.
You can listen or watch here.
The Method of Exhaustion
New video up on YouTube today, covering a centuries-old calculus technique called the Method of Exhaustion. The technique is still relevant today as a stepping stone to understanding how modern calculus works.
The YouTube playlist for my "Calculus for Machine Learning" course is here.
My Favorite Books
I ask every guest on the SuperDataScience show for a book recommendation. Now it's my turn! In this post, I discuss what I love about my favorites in both the fiction and non-fiction realms 📚
Read MoreIntro to Integral Calculus
Today’s video is a quick intro to Integral Calculus, the other branch of the mathematical field alongside Differential Calculus (which was introduced in the preceding video, released on Monday).
The YouTube playlist for my "Calculus for Machine Learning" course is here.