简体中文
EnglishWho the Jingles (Hong Jing) Site Is For, and How a New Reader Should Start
Updated
The site is most useful for machine learning learners and early-career practitioners who want plain-language explainers, plus readers curious about where ML meets neuroscience. A new reader should start with the beginner-oriented posts — "7 ways to get started into ML, ranked from easiest to hardest" and "How Convolutional Layers Work in Deep Learning Neural Networks?" — then use the tag index or the blog archive (the listing runs across at least three pages) to follow a specific thread. It is a personal blog by Hong Jing (Jingles), described on the site as a data scientist and Alibaba PhD student who writes about machine learning and neuroscience, so expect a mix of introductory explainers and more advanced or research-flavored pieces rather than a structured course.
Who gets the most from it
| Reader | What the site offers | Suggested entry point |
|---|---|---|
| ML beginner | Ranked paths into the field, conceptual explainers on neural network building blocks | "7 ways to get started into ML" then the convolutional layers explainer |
| Practitioner filling gaps | Explainers on optimizers, transformers, sequence problems, evaluation | "Why Learned Optimizers Outperform Standard Optimizers like Adam," "Illustrated Guide to Transformer" |
| Neuroscience-curious reader | Posts on brain signals, AI and neuroscience, and how the two fields relate | "The Fascinating Relationship between AI and Neuroscience," "Decoding Brain Signals with Machine Learning and Neuroscience" |
| Hands-on tinkerer | Project write-ups and quick builds | "Realistic Deepfakes in 5 Minutes on Colab," the Overcooked multi-agent RL post |
The content spans a wide range: from "Babies are awesome... Humans are the OG neural net" style short reflections to technical pieces on state-of-the-art language models and 3D object understanding. That range is the main thing to plan around — it is not sequenced, so you choose by topic rather than by level.
A practical reading order
- Orient yourself. Read the "7 ways to get started into ML, ranked from easiest to hardest" post. It is explicitly ordered by difficulty, which makes it the best single map of the site's beginner material.
- Build the fundamentals. Move to "How Convolutional Layers Work in Deep Learning Neural Networks?" and "An Introduction to Signals." These are conceptual explainers, so read them for the mechanism, not for runnable code.
- Pick a thread that matches your goal. For model architecture, go to the transformer guide and the sequence prediction post. For training and evaluation, use the learned optimizers and "Intelligent AI? Here's How You Can Evaluate" pieces.
- Cross into neuroscience if that is your interest. The AI-and-neuroscience and brain-signal decoding posts are the natural pairing with the ML material.
- Use the archive and tags to go deeper. The blog listing is paginated (pages 1–3 are visible), and there is a "Browse articles by tags" entry, which is the fastest way to find everything on one subject.
Where to find projects and profiles
The site's navigation includes Blog, Showcase, Projects, Published Work, and About, alongside links to Medium, LinkedIn, and GitHub. If you want code or ongoing work rather than articles, the Projects and Showcase sections and the GitHub link are the places to look; the About page is the source for background details.
Common snags
- Expecting a course. The posts are independent articles, so there are gaps between topics; fill them with a structured course elsewhere and use this site for explanation and inspiration.
- Expecting runnable notebooks everywhere. Most pieces are explanatory. The Colab deepfakes post is the clearest hands-on example in the listing; treat the rest as reading.
- Judging level from titles alone. A title like "State-of-the-Art Language Models in 2020" is dated by its year, so check whether a post still reflects current practice before relying on it.