Skip to content
DevMeme
4638 of 7590
AI ML Post #5086 · source on Telegram

Deep-learning devotee calls statistician brainwashed, statistician responds with simple linear model

Description

The meme has a light blue background and shows two minimalist black figures facing each other. On the left, a plain silhouette is surrounded by AI/ML visuals: the OpenAI logo, TensorFlow and PyTorch logos, a colorful JAX-style icon, the scikit-learn logo, a convolutional-network block diagram, a decision-tree flowchart, a non-linear decision-boundary scatter plot, and a small feed-forward neural-network sketch. A large white speech bubble above this figure reads, in bold black letters, "they brainwashed you". On the right stands a bearded Wojak-style character in a black hoodie; a smaller bubble above him says "really?", and he holds a white sign with the linear regression formula "Y = Xβ + ε". The juxtaposition humorously contrasts modern deep-learning hype and framework fandom with classical statistics, poking fun at claims that sticking to simple models means being "brainwashed."

Comments

12
Anonymous ★ Top Pick Call me brainwashed all you like - my linear regression ships with a git push; your 3-billion-parameter sandbox needs a Kubeflow séance every time the feature store hiccups
  1. Anonymous ★ Top Pick

    Call me brainwashed all you like - my linear regression ships with a git push; your 3-billion-parameter sandbox needs a Kubeflow séance every time the feature store hiccups

  2. Anonymous

    After 15 years of watching ML frameworks come and go, you realize the real deep learning was discovering that your stakeholders just needed a pivot table all along - but hey, at least your GPU farm is keeping the office warm

  3. Anonymous

    When your junior ML engineer insists on deploying a 50-layer transformer with attention mechanisms for a problem that's literally Y = Xβ + ε, but you can't argue because they got 0.001% better accuracy on the validation set and now it takes 47 GPUs and three days to retrain

  4. Anonymous

    Brainwashed? No - it’s just preprocessing: standardization, dropout, and L2 on my priors. You’re the one still overfitting to Xβ+ε

  5. Anonymous

    Deep learning: linear regression that convinced VCs to fund GPU farms for the epsilon term

  6. Anonymous

    Six months of GPU burn later, the transformer beat OLS by 0.2% - right after we forgot to shuffle; turns out the strongest regularizer in enterprise ML is the marketing budget

  7. @Vlasoov 3y

    Based

  8. Deleted Account 3y

    Explain

    1. @deathstranger97 3y

      https://www.simplilearn.com/what-is-multiple-linear-regression-in-machine-learning-article

      1. @callofvoid0 3y

        ah another inaccessable source of knowledge

  9. @im_ali_pj 3y

    based😂

  10. @Draxly 3y

    so called

Use J and K for navigation