hamasho

I'm not an expert in machine learning, but rather a web developer and data engineer helping develop a system to detect diseases from endoscopy images using the model developed by other ML engineers. And it was 5 years ago when I worked on the project, so please take it with a grain of salt.

If you want to learn machine learning for healthcare in general, it may help to start problems with tabular data like CSVs instead of images. Image processing is a lot harder, and takes a lot of time and computational power. But it's best to learn what you're interested in the most.

Anyway, first you need to be familiar with basic; Python, machine learning, and popular libraries like scikit-learn, matplotlib, numpy, and pandas. Those are tons of articles, textbooks, and videos to help you learn them.

If you grasp the basics, I think it's better to learn from actual code to train/evaluate models rather than more theories. Kaggle may be a good starting point. They host a lot of competitions for machine learning problems. There are easy competitions for beginners, and competitions and datasets in the medical field.

You can view notebooks (actual code to solve those problems well written by experts) and popular ones are very educational. You can learn a lot by reading those code, understanding concepts and how to use libraries, and modifying some code to see how it changes the result. ChatGPT is also helpful.

If you want to learn image classification, the technology used to detect objects from images and videos is called image classification and object detection. It uses CNN, one of the deep neural networks. You also need to learn basic image processing, how to train a deep neural network, how to evaluate, and libraries like OpenCV/Pillow/PyTorch/TorchVision. There are a lot of image classification competitions in the medical field on Kaggle too[0][1].

To run those notebooks, I recommend Google Colab. Image processing often uses a lot of GPUs, and you may not have GPUs, or even if you have it's difficult to set up the right environment. It's easier to use those dedicated cloud services and it doesn't cost much.

It's hard to learn, but sometimes fun, so enjoy your journey!

[0] https://www.kaggle.com/datasets/paultimothymooney/chest-xray... [1] https://www.kaggle.com/code/arkapravagupta/endoscopy-multicl...

I'm not familiar with the medical field. To what extent does experience using mice translate to humans?

I worked on a project to analyze endoscope videos to find diseases. I examined a lot of images and videos annotated with symptoms of various diseases labeled by assistants and doctors. Most of them are really obvious, but others are almost impossible to detect. In rare cases, despite my best efforts, I couldn't see any difference between the spot labeled as a symptom of cancer and the surrounding area. There's no a-ha moment, like finding an insect mimicking its environment. No matter how many times I tried, I just couldn't see any difference.

This reminds me of Korea's progressive feminist hand gesture [0]. For some reason, a video about Korea's gender war [1] ended up in my YouTube recommendation and somehow I decided to watch the 47-minute video (and another 110 minutes for part 2)...

Basically, aggressive feminist groups use a hand gesture as a disrespect against men (or misogynists), but the gesture is so general and anti-feminism is so large in Korea, that a lot of people mistook the otherwise normal picture in anime/games as a hidden attack against men. It caused riots and several people lost their jobs or sometimes the entire projects/companies went down.

I think it's not a good idea to associate a very natural gesture with horrible intentions...

[0] https://edition.cnn.com/2021/10/02/business/south-korea-busi... [1] https://www.youtube.com/watch?v=-Im4YAMWK74

Thanks for the tip! I'll try leaving a task unfinished to start the next day with a small win. As a remote dev, I struggle to be productive all day. My company trusts me and gives me lots of flexibility, which sometimes allows me to procrastinate for days, leading to self-blaming and anxiety. I know starting the day in a good mood makes the rest of the day easier. Kicking off with a small victory definitely helps my productivity!

Oh, thanks for pointing out.

I didn't know what entropy means in software, so here's the definition[0]:

----

  Software entropy is a measure of the disorder or complexity of a software system. It is a natural tendency for software entropy to increase over time, as new features are added and the codebase becomes more complex.

  High entropy in software development means that the code is difficult to understand, maintain, and extend. It is often characterized by:

      Duplicated code: The same code or functionality is repeated in multiple places, which can make it difficult to find and fix bugs.
      Complex logic: The code is difficult to follow and understand, which can make it difficult to add new features or fix bugs without introducing new ones.
      Poor documentation: The code is not well-documented, which can make it difficult for new developers to understand and contribute to the codebase.
      Technical debt: The code has been patched and modified over time without proper refactoring, which can lead to a tangled and cluttered codebase.

  Low entropy in software development means that the code is well-organized, easy to understand, and maintain. It is often characterized by:

      Well-designed architecture: The code is structured in a logical way, with clear separation of concerns.
      Consistent coding style: The code follows a consistent coding style, which makes it easy to read and understand.
      Comprehensive documentation: The code is well-documented, with clear explanations of the code's purpose and functionality.
      Minimal technical debt: The code has been refactored regularly to remove technical debt, which makes it easy to add new features and fix bugs without introducing new ones.

[0] https://www.kisphp.com/python/high-and-low-entropy-in-softwa...

For me, it's not just social media but also Hacker News.

I often open Hacker News, scan the titles, and open all interesting threads in separate tabs for later reading (I opened this thread the same way). Each thread usually contains insightful discussions and useful references on topics I enjoy, like music, programming, and recently, LLMs. Unlike Reddit, I often read the original articles instead of just shit-posting on clickbait titles.

Most of the discussions and articles are information-dense, requiring a significant amount of time and mental effort to read through them all (and close all the tabs). It's not uncommon for these dozens of open tabs to drain my willpower just by being there all day long.

I feel like I'm driven by FOMO rather than sheer curiosity. Maybe some information diet helps?

I think if you know the domain well, it's not "premature" at all.

I agree. I think many people confuse LLM abilities because LLMs are so similar to humans that users expect them to do everything humans can do. Remembering what was said is a basic human ability, so it might surprise some users that LLMs can’t do that (although ChatGPT has recently implemented a memory feature).

I get that it’s usually seen as bad practice to write down all the points and read from the slides, but honestly, I like that style sometimes. My monkey brain gets distracted easily, and I often lose track. Having detailed agendas displayed on the screen helps me follow along better.

Google is abusing its monopoly power, so I'm glad other companies are joining the search market. But we need a new framework to reward authors of primary information fed to LLM.

I often google Reddit posts to find products or services. Posts with a lot of comments usually have better solutions. I also check out YouTube reviews, especially ones with few dislikes, using addons like "Return YouTube Dislike." These methods are easy and pretty reliable.

But it means the platform will soon be flooded with bot-generated spam.

Not GP, but navigating to an unrelated scammy site just having a link to the actual site is a terrible and unethical job by Google. Imagine if you search "youtube" and the top result is not YouTube but some scammy site just having a link to YouTube. It's not about click counts, if the youtube downloader has bad UX and requires extra clicks, it's a bit inconvenient but ok.

John Oliver explained this topic well. I recommend to watch the video.

https://youtu.be/jtxew5XUVbQ?si=oYLWIhuD0srg13hN

It may be summarized to what they want, but it can be what I want too. All YouTube recommendations are already what I want to see. Videos from creators who share my world view, criticizing what I hate, and praising what I've already liked. If this Chrome's new feature is also personalized in this manner, I can read articles against my view and still get only what I want. My world view will become increasingly polarized, and it's already quite biased.

I like a package management system integrated with project management, such as JS's package.json or Python's pyproject.toml. I want to manage project scripts, tool configs, and dependencies in one place. It's sometimes annoying, especially for larger projects, but overall, I hope more languages adopt that style.

They're Amish in software world.

Simple, instant, and elegant. I hope search works in the same way. Life would be much easier if I started typing and a bunch of answers appeared immediately. I don't have to visit search results one by one just to find out it's a curation site of stack overflow.

Off topic but Jupyter cells on GitHub can't display horizontally long content and it frustrates me a lot. This small piece of code for the browser console helps me see more content, but it only works in large displays.

    $("[data-type='ipynb']").style.width = '100%'

Thanks! It's reasonable. But I'm pretty sure yours is more complex than many typical SaaS or CRUD-based websites.

If you don't mind, can I ask why you use plain JavaScript rather than TypeScript?

The product is truly remarkable, so what technologies used under the hood doesn't matter, but I think TypeScript is natural choice, especially if you 1) start a project from scratch in 2023, 2) have complex local states like choosing equipment and body parts, and proceeding exercises step by step until completed in a single page, and 3) have complex data structure (like exercises, histories, and seamless connection between locally stored data and remotely stored data used only after registering).

edit: typo

This is truly fantastic, thank you!

The overall experiences are so great. And I have strong respect for apps like this where I can enjoy a lot of features without being required to create an account. I understand it's difficult in corporations where KPIs like registration rate matters. But even when given a choice, it must be challenging to design seamless UX/UI and write code which works well with both locally stored data (without an account) and remotely stored data (with an account). I hope more apps adopt this approach. There are very few examples. Excalidraw is the one that comes to mind.

Anyway I really love your app, thank you! I'll use this in the next gym session (if I finally hit the gym...)

I'm concerned that the progress of LLMs will enable more bureaucratic and BS in corporate cultures.

OpenAI's CEO tweeted [0] that people using ChatGPT to compose a polite, business grade email from bullet points, and then the receivers decode it into bullet points. If this is the future, I guess we may see more verbose, a little bit cryptic sentences in emails (just like compiled code is getting more cryptic for human). And we already have lots of meetings that most people think meaningless. Now the records of meetings are automatically transcribed, stored in company's knowledge base, and retrieved by simple chats with ChatGPT. It gives a false impression of productivity, and encourages more and more meetings!

I think we need to refactor the organization/culture first, or we'll engage more complicated business ritual and pay small amount to OpenAI every time we need to reduce the BS that we've created unnecessarily.

[0] https://twitter.com/sama/status/1631394688384270336?lang=en

I'm pretty sure this is what my chatgpt thinks whenever I say hi. I know it's annoying, but you respond so quickly and gladly, so I can't stop starting the chat with hi. I should stop this, or I would begin to do the same to real humans on Slack.

I've finally come to the conclusion that user tracking is generally a poor practices and should be regulated.

As a web developer, I didn't see it as a big problem. We always do it to maximize ad revenue, find out where users leave to increase conversion rate, and simply to improve UX. But even when the intent is to improve UX, tracking is inappripriate.

Imagine if a robot vacuum recorded videos of your home and uploaded them so that bunch of ML engineers can see and use it to improve the algorithm. Or the videos of your car's camera (both of inside and outside). I mean, I'm not surprised if this is already happening, but it's a disturbing thought and should be regulated.

We can certainly develop functional services without tracking users.

I realized how long I watched YouTube videos, I don't learn any new knowledge. I can learn a little about things I already know enough, or I can grasp overview of new topics, but it's never enough to use in actual conversations.

The worst part is the false sense of improvement. IMO, Learning nothing, yet feeling a sense of improvement is worse than learning nothing at all.

This AI bull run will make people like Son and Cathie Wood earn much money. Nothing is wrong with that but I hate those “Why don't you invest in TQQQ are you an idiot!?” people dominate everywhere.

I thought this was something similar to Microsoft mandating membership to update OS

This comment makes me wonder why ad companies haven't already used this to show ads. They are notorious for using any technology to grab users' attentions.