Close Menu
    Trending
    • OpenAIs nya webbläsare ChatGPT Atlas
    • Creating AI that matters | MIT News
    • Scaling Recommender Transformers to a Billion Parameters
    • Hidden Gems in NumPy: 7 Functions Every Data Scientist Should Know
    • Is RAG Dead? The Rise of Context Engineering and Semantic Layers for Agentic AI
    • ChatGPT Gets More Personal. Is Society Ready for It?
    • Why the Future Is Human + Machine
    • Why AI Is Widening the Gap Between Top Talent and Everyone Else
    ProfitlyAI
    • Home
    • Latest News
    • AI Technology
    • Latest AI Innovations
    • AI Tools & Technologies
    • Artificial Intelligence
    ProfitlyAI
    Home » Things I Wish I Had Known Before Starting ML
    Artificial Intelligence

    Things I Wish I Had Known Before Starting ML

    ProfitlyAIBy ProfitlyAIJuly 23, 2025No Comments9 Mins Read
    Share Facebook Twitter Pinterest LinkedIn Tumblr Reddit Telegram Email
    Share
    Facebook Twitter LinkedIn Pinterest Email


    Ahh, the ocean.

    trip on the Mediterranean Sea, I discovered myself mendacity on the seaside, staring into the waves. Girl Luck was having day: the solar glared down from a blue and cloudless sky, heating the sand and salty sea round me. For the primary time shortly, I had downtime. There was nothing associated to ML within the distant area the place I used to be, the place the tough roads would have scared away anyone who’s used to the even pavements of western nations.

    Then, away from work and, partially, civilization, someplace between zoning out and full-on daydreaming, my ideas started to float. In our day-to-day enterprise, we’re too, properly, busy to spend time doing nothing. However nothing is powerful phrase right here: as my ideas drifted, I first recalled current occasions, then contemplated about work, after which, finally, arrived at machine studying.

    Perhaps traces of my earlier article—where I reflected on 6.5 years of “doing” ML—had been nonetheless lingering behind my thoughts. Or perhaps it was merely the entire absence of something technical round me, the place the ocean was my solely companion. Regardless of the cause was, I mentally began rehearsing the years behind me. What had gone properly? What had gone sideways? And—most significantly—what do I want somebody had informed me at first?

    This submit is a group of these issues. It’s not meant to be a listing of dumb errors that I urge others to keep away from in any respect prices. As a substitute, it’s my try to jot down down the issues that might have made my journey a bit smoother (however solely a bit, uncertainty is important to make the longer term simply that: the future). Elements of my listing overlap with my earlier submit, and for good cause: some classes are value repeating, and studying once more.

    Right here’s Half 1 of that listing. Half 2 is presently buried in my sandy, sea-water stained pocket book. My plan is to comply with up with it within the subsequent couple of weeks, as soon as I’ve sufficient time to show it into a high quality article.

    1. Doing ML Principally Means Making ready Knowledge

    This can be a level I strive to not suppose an excessive amount of about, or it should inform me: you didn’t do your homework.

    Once I began out, my inside monologue was one thing like: “I simply wish to do ML.” No matter that meant; I had visions of plugging neural networks collectively, combining strategies, and working large-scale coaching. Whereas I did all of that at one level or one other, I discovered that “doing ML” typically means spending numerous time simply making ready the info so that you could finally prepare a machine studying mannequin. Mannequin coaching, paradoxically, is usually the shortest and final a part of the entire course of.

    Thus, each time I lastly get to the mannequin coaching step, I mentally breathe a sigh of aid, as a result of it means I’ve made it by way of the invisible half: making ready the info. There’s nothing “sellable” in merely making ready the info. In my expertise, making ready the info will not be noticeable in any method (so long as it’s executed properly sufficient).

    Right here’s the standard sample for it:

    • You’ve got a undertaking.
    • You get a real-world dataset. (Should you work with a well-curated benchmark dataset, then you definately’re fortunate!)
    • You wish to prepare a mannequin.
    • However first… knowledge cleansing, fixing, merging, validating.

    Let me provide you with a private instance, one which I’ve told as a funny story (which it’s now. Again then, it meant redoing a couple of days of machine studying work beneath time strain…).

    I as soon as labored on a undertaking the place I wished to foretell vegetation density (utilizing the NDVI index) from ERA5 climate knowledge. ERA5 is an enormous gridded dataset, freely obtainable from the European Centre for Medium-Vary Climate Forecasts. I merged this dataset with NDVI satellite tv for pc knowledge from NOAA (principally, the American climate company), fastidiously aligned the resolutions, and the whole lot appeared fantastic—no form mismatches, no errors had been thrown.

    Then, I known as the info preparation executed and skilled a Imaginative and prescient Transformer mannequin on the mixed dataset. A couple of days later, I visualized the outcomes and… shock! The mannequin thought Earth was the wrong way up. Actually—my enter knowledge was right-side up, however the goal vegetation density was flipped on the equator.

    What had occurred? A delicate bug in my decision translation flipped the latitude orientation of the vegetation knowledge. I hadn’t seen it as a result of I used to be spending numerous time on knowledge preparation already, and wished to get to the “enjoyable half” shortly.

    This sort of mistake hones in an vital level: real-world ML initiatives are knowledge initiatives. Particularly exterior educational analysis, you’re not working with CIFAR or ImageNet. You’re working with messy, incomplete, partially labellel, multi-source datasets that require:

    • Cleansing
    • Aligning
    • Normalizing
    • Debugging
    • Visible inspection

    And much more, that listing is non-exhaustive. Then repeating all the above.

    Getting the info proper is the work. Every part else builds on that (sadly invisible) basis.

    2. Writing Papers Is Like Making ready a Gross sales Pitch

    Some papers simply learn properly. You won’t be capable to clarify why, however they’ve a movement, a logic, a readability that’s onerous to disregard. That’s not often by chance*. For me, it turned out that writing papers resembles crafting a really particular form of gross sales pitch. You’re promoting your thought, your method, your perception to a skeptical viewers.

    This was a shocking realization for me.

    Once I began out, I assumed most papers appeared and felt the identical. All of them had been “scientific writing” to me. However over time, as I learn extra papers I started to note the variations. It’s like that saying: to outsiders, all sheep look the identical; to the shepherd, every one is distinct.

    For instance, evaluate these two papers that I got here throughout lately:

    Each use machine studying. However they communicate to totally different audiences, with totally different ranges of abstraction, totally different narrative kinds, and even totally different motivations. The primary one assumes that technical novelty is central. The second focuses on relevance for purposes. Clearly, there is also the visible distinction between the 2.

    The extra papers you learn, the extra you understand: there’s not one technique to write a “good” paper. There are a lot of methods, and the best way varies relying on the viewers.

    And until you’re a kind of very uncommon good minds (suppose Terence Tao or somebody of that caliber), you’ll probably want help to jot down properly. Particularly when tailoring a paper for a particular convention or journal. In apply, which means working carefully with a senior ML one that understands the sphere.

    Crafting paper is like making ready a gross sales pitch. You could:

    • Body the issue the correct method
    • Perceive your viewers (i.e. goal venue)
    • Emphasize the elements that resonate most
    • And polish till the message sticks

    3. Bug Fixing Is the Manner Ahead

    Years in the past, I had that romantic thought of ML as exploring elegant fashions, inventing new activation capabilities, or crafting intelligent loss capabilities. That could be true for a small set of researchers. However for me, progress typically appeared like: “Why doesn’t this code run?”. Or, much more irritating: “That code simply ran a couple of seconds ago-why does it not run now?”

    Let’s say your undertaking requires utilizing Imaginative and prescient Transformers on environmental satellite tv for pc knowledge (i.e., the mannequin aspect of Part 1 above). You’ve got two choices:

    1. Implement the whole lot from scratch (not advisable until you’re feeling significantly adventurous, or must do it for course credit).
    2. Discover an current implementation and adapt it.

    In 99% of the instances, possibility 2 is the apparent selection. However “simply plug in your knowledge” virtually by no means works. You’ll run into:

    • Totally different compute environments
    • Assumptions about enter shapes
    • Preprocessing quirks (corresponding to knowledge normalization)
    • Laborious-coded dependencies (of which I’m responsible, too)

    Shortly, your day can develop into an infinite sequence of debugging, backtracking, testing edge instances, modifying dataloaders, checking GPU reminiscence**, and rerunning scripts. Then, slowly, issues start to work. Ultimately, your mannequin trains.

    However it’s not quick. It’s bug fixing your method ahead.

    4. I (Very Actually) Gained’t Make That Breakthrough

    You’ve positively heard of them. The Transformer paper. The GANs. Steady Diffusion. There’s a small half in my that thinks: perhaps I’ll be the one to jot down the subsequent transformative paper. And certain, somebody has to. However statistically, it most likely gained’t be me. Otherwise you, apologies. And that’s fantastic.

    The works that trigger a area to vary quickly are distinctive by definition. These works being distinctive straight implies that almost all works, even good work, are barely acknowledged. Generally, I nonetheless hope that certainly one of my initiatives would “blow up.” However, up to now, most didn’t. Some didn’t even get revealed. However, hey, that’s not failure—it’s the baseline. Should you anticipate each paper to be a house run, then you’re on the quick lane to disappointment.

    Closing ideas

    To me, Machine studying typically seems as a smooth, cutting-edge area—one the place breakthroughs are simply across the nook and the place the “doing” means good folks make magic with GPUs and math. However in my day-to-day work, it’s not often like that.

    Extra typically, my day-to-day work consists of:

    • Dealing with messy datasets
    • Debugging code pulled from GitHub
    • Redrafting papers, time and again
    • Not producing novel outcomes, once more

    And that’s okay.


    Footnotes

    The earlier article talked about: https://towardsdatascience.com/lessons-learned-after-6-5-years-of-machine-learning/

    * In case you are , my favourite paper is that this one: https://arxiv.org/abs/2103.09762. I learn it one 12 months in the past on a Friday afternoon.

    ** To at the present time, I nonetheless get mail notifications about how clearing the GPU reminiscence is unattainable in TensorFlow. This 5-year old GitHub issue offers the small print.



    Source link

    Share. Facebook Twitter Pinterest LinkedIn Tumblr Email
    Previous ArticleNumPy API on a GPU?
    Next Article Google DeepMind’s new AI can help historians understand ancient Latin inscriptions
    ProfitlyAI
    • Website

    Related Posts

    Artificial Intelligence

    Creating AI that matters | MIT News

    October 21, 2025
    Artificial Intelligence

    Scaling Recommender Transformers to a Billion Parameters

    October 21, 2025
    Artificial Intelligence

    Hidden Gems in NumPy: 7 Functions Every Data Scientist Should Know

    October 21, 2025
    Add A Comment
    Leave A Reply Cancel Reply

    Top Posts

    The Dangers of Deceptive Data Part 2–Base Proportions and Bad Statistics

    May 9, 2025

    What We Need to Know About AI in Emotion Recognition in 2024

    April 5, 2025

    Pope Leo XIV Declares AI a Threat to Human Dignity and Workers’ Rights

    May 12, 2025

    A Gentle Introduction to Backtracking

    June 30, 2025

    Data Drift Is Not the Actual Problem: Your Monitoring Strategy Is

    June 4, 2025
    Categories
    • AI Technology
    • AI Tools & Technologies
    • Artificial Intelligence
    • Latest AI Innovations
    • Latest News
    Most Popular

    Building Video Game Recommender Systems with FastAPI, PostgreSQL, and Render: Part 1

    September 25, 2025

    AI system predicts protein fragments that can bind to or inhibit a target | MIT News

    April 5, 2025

    Google’s New AI Mode Could Replace How You Search, Shop, and Travel Forever

    May 2, 2025
    Our Picks

    OpenAIs nya webbläsare ChatGPT Atlas

    October 22, 2025

    Creating AI that matters | MIT News

    October 21, 2025

    Scaling Recommender Transformers to a Billion Parameters

    October 21, 2025
    Categories
    • AI Technology
    • AI Tools & Technologies
    • Artificial Intelligence
    • Latest AI Innovations
    • Latest News
    • Privacy Policy
    • Disclaimer
    • Terms and Conditions
    • About us
    • Contact us
    Copyright © 2025 ProfitlyAI All Rights Reserved.

    Type above and press Enter to search. Press Esc to cancel.