The Lindahl Letter: On Machine Learning

By Nels Lindahl

Greetings, inspired reader of machine learning glory. If you made it here, then thank you for the consideration. This weighty tome of my insights covers my weekly thoughts on machine learning. Pretty much everybody seemed to be writing a Substack when I began this endeavor at the start of 2021. It seemed like a great time to start writing a regular weekly publication. You should be able to find fresh content every Friday throughout for the foreseeable future online. I expect to be publishing a post related to machine learning, artificial intelligence, or business strategy. This manuscript you are reading the description for right now contains all my 2021 missives in one curated collection for hours of reading joy. Assuming you have the digital version of this the links should be easy to click and are all shared as footnotes within the chapters. If you have a physical copy of this book which by the way I appreciate that you have, then you will be required to type the links out into an internet browser manually. My series of weekly posts was compiled into a manuscript. Each post has been edited from the original form into a more publication friendly format. Substack as a platform provides a lot of freedom to include links and embedded content that does not translate into the written page.

• Language: English
• Publisher: Lulu.com
• Release date: Jan 9, 2022
• ISBN: 9781716017407

Book preview

Dedication

To those who look at what could be and thread the needle to get there…

Acknowledgments

Thanks to all of my first-year Substack post readers who helped inspire this effort. It was great to get feedback along the way while writing a book. I’m thinking next year this method will be a great way to do the same thing again to produce a manuscript for 2022.

Table of Contents

Dedication

Acknowledgments

Preface

Substack Week 1: Machine learning return on investment

Substack Week 2: Machine learning frameworks & pipelines

Substack Week 3: Machine learning teams

1. Where does the talent come from?

2. How do you get the talent to work together?

Substack Week 4: Have a machine learning strategy…revisited

1. What exactly is a machine learning strategy?

2. Do you even need a machine learning strategy?

Substack Week 5: Let your ROI drive a fact-based decision-making process

Substack Week 6: Understand the ongoing cost and success criteria as part of your machine learning strategy

Substack Week 7: Plan to grow based on successful ROI

Substack Week 8: Is the machine learning we need everywhere now?

Substack Week 9: Valuing machine learning use cases based on scale

Substack Week 10: Model extensibility for few-shot GPT-2

Substack Week 11: What is machine learning scale? The where and the when of machine learning usage

Brief presentation abstract

Research disclaimer

Section 1: Initiating: What is scale exactly?

Selection 2: Analyzing: Common machine learning use cases: Scale versus maturity

Section 3: Directing: Leaning into machine learning scale

Substack Week 12: Confounding within multiple machine learning model deployments

Substack Week 13: Building out your MLOps

What did we cover?

Substack Week 14: My Ai4 Healthcare NYC 2019 talk revisited

1. Where does the talent come from?

2. How do you get the talent to work together?

3. What are these workflows, and why do they matter?

4. What problems can you solve with machine learning?

5. What exactly is a machine learning strategy?

6. What do you mean by machine learning vectors?

7. What is a compendium of KPIs?

8. What are some examples of machine learning turning the wheel?

Substack Week 15: What are people really doing with machine learning?

Substack Week 16: Ongoing machine learning cloud costs

Bucket 1: Things you can call, such as external API services

Bucket 2: Places you can be, or ecosystems where you can build out your footprint

Bucket 3: Building something yourself, such as open-source and self-tooled solutions

Substack Week 17: Figuring out machine learning readiness

Substack Week 18: Could machine learning predict the lottery?

Substack Week 19: Fear of missing out on machine learning

Substack Week 20: Week 20 Lindahl Letter recap edition

Substack Week 21: Doing machine learning work

Substack Week 22: Machine learning graphics

Substack Week 23: Fairness and machine learning

Substack Week 24: Evaluating machine learning

Substack Week 25: Teaching kids

Substack Week 26: Machine learning as a service

Substack Week 27: The future of machine learning

Substack Week 28: Machine learning certifications?

Substack Week 29: Machine learning feature selection

Substack Week 30: Integrations and your machine learning layer

Substack Week 31: Edge machine learning integrations

Substack Week 32: Federating your machine learning models

Substack Week 33: Where are artificial intelligence investments coming from?

Substack Week 34: Where are the main artificial intelligence labs?

Substack Week 35: Explainability in modern machine learning

Substack Week 36: AIOps/MLOps: Consumption of artificial intelligence services versus operations

Substack Week 37: Reverse engineering GPT-2 or GPT-3

Substack Week 38: Do most machine learning projects fail?

Substack Week 39: Machine learning security

Substack Week 40: Applied machine learning skills

Substack Week 41: Machine learning and the metaverse

Substack Week 42: Time crystals and machine learning

Substack Week 43: Practical machine learning

Substack Week 44: Machine learning salaries

Substack Week 45: Prompt engineering and machine learning

Substack Week 46: Machine learning and deep learning

Substack Week 47: Anomaly detection and machine learning

Substack Week 48: Machine learning applications revisited

Substack Week 49: Machine learning assets

Postlogue

About the author

Preface

Greetings, inspired reader of machine learning glory.

Pretty much everybody seemed to be writing a Substack when I began this endeavor at the start of 2021. As a platform, Substack grew very rapidly, capturing a few moments of attention from the public mind. Currently only a few Substack authors still stand out from the crowd. Like all newsletters or writing efforts with a community, the audience makes or breaks ongoing success. This series of weekly posts was compiled into a manuscript. Each post has been edited from the original form into a more publication-friendly format. Substack as a platform provides a lot of freedom to include links and embedded content, which does not translate into the written page. 

One of the things I did notice is that the first few posts were much longer than the ones at the end of the series. It appears that when I started talking about machine learning in general, the amount of content related to things I wanted to say was much larger. Given that each week is written to be independently consumed, all of my references are handled in a straightforward footnote method. Any aside, link, or acknowledgment to another author happens in the footnotes ending the chapter.

Things in this highly technology-driven space are changing rapidly. I included the date of publication as a frame of reference.

Dr. Nels Lindahl

Broomfield, Colorado

December 15, 2021 @ 7:49 PM

Substack Week 1: Machine learning return on investment

Published on January 29, 2021

Be strategic with your machine learning efforts.

Be

strategic

with

your

machine

learning

efforts.

Seriously, those seven words should guide your next steps along the machine learning journey. Take a moment and let that direction (strong guidance) sink in and reflect on what it really means for your organization. You have to take a moment and work backward from building strategic value for your organization to the actual machine learning effort you are undertaking. Inside that effort you will quickly discover that operationalizing machine learning efforts to generate strategic value will rely on a solid plan for return on investment (ROI). Make sure you are beginning with that end in mind to increase your chances of success. Taking actions within an organization of any kind at the scale machine learning is capable of delivering, without understanding the potential ROI or potential loss, is highly questionable. That is why you have to be strategic with your machine learning efforts from start to finish.   

You have to set up and run a machine learning strategy from the top down. Executive leadership has to understand and be invested in guiding things toward the right path (a truly strategic path) from the start. Start by making an effort to begin with a solid strategy in the machine learning space. It might sound harder than it is in practice. You don’t need a complicated center of excellence or massive investment to develop a strategy. Your strategy just needs to be linked to the budget and ideally to a budget key performance indicator (KPI). Every budget results in the process of spending precious funds, and keeping a solid KPI around machine learning ROI levels will help ensure your strategy ends on a strong financial footing for years to come. All spending of an organization's precious resources should translate to a KPI of some type. That is how your results will let you confirm that the funding is being spent well and that solid decision-making is occurring. You have to really focus and ensure that all spending is tied to that framework when you operationalize the organization's strategic vision to be aligned financially to the budget. 

That means that the machine learning strategy you are investing in has to be driven to achieve a certain ROI tied directly to solid budget-level KPIs. You might feel like that line has been repeated. If you noticed that repetition, then you are paying attention and well on your way to future success. Reading comprehension goes a long way to translating written argument to action. That KPI-related tieback you are creating is only going to happen with a solid machine learning strategy in place. It has to be based on prioritizing and planning for ROI. Your machine learning pipelines and frameworks have to be aligned toward that goal. That is ultimately the cornerstone of a solid strategic plan when it comes to implementing machine learning as part of a long-term strategy.

We are about 500 words into this book, and it might be time to simply recap the message being delivered so far. Be ready to do things in a definable and repeatable way. Part of executing a strategy with quality is doing things in a definable and repeatable way. That is the essence of where quality comes from. You have to know what plan is being executed and focus on and support the plan in ways that make it successful at your desired run rate. In terms of deploying machine learning efforts within an enterprise, you have to figure out how the technology is going to be set up and invested in and how that investment is going to translate to use cases with the right ROI.  

Know the business value for the use case instead of letting solutions chase problems. Just because you can do a thing does not always mean that you should. Having the ability to deploy a technology does create the potential of letting a technology-based solution chase a problem. Building up technology for machine learning in a very theoretical and lab-based way and then chasing use cases is a terrible way to accidentally stumble on an ROI model that works. The better way forward is to know the use cases and have a solid strategy to apply your technology. That means finding the right machine learning frameworks and pipelines to support your use cases in powerful ways across the entire organization. 

This is a time to be planful. Right here, right now, in this moment of consideration you can elect to be planful going forward. Technology for machine learning is becoming increasingly available and plentiful. No code, low code, and just solidly integrated solutions are becoming omnipresent in the technology landscape. Teams from all over the organization are probably wanting to try proof of concepts, and vendors are bringing in a variety of options. People are always ready to pitch the value of machine learning to the organization. Both internal and external options are plentiful. It is an amazing time for applied machine learning. You can get into the game in a