4 ms·
Show HN: Deep Learning in TensorFlow – The Roadmap for Study and Learning
- mywrathacademia 8y agoThis guide will be useful.
- partingshots 8y agoNot disparaging the author, good on you for working on building this! I’m sure you learned a lot from just compiling everything together. My question is, isn’t everything in this guide pretty much just a straight up copy of the actual TensorFlow docs/guides? What’s the difference?
- yeonsh 8y agoThanks for the list.
- eternal_virgin 8y agoIt would be great to have some sort of roadmap depending on "what" the user is trying to do. i.e. if I was a total noob and I wanted to make an application using AI to detect if people in the crowd were bored, I would have no idea where to start without reading/researching for hours online on different fields and models that work and how they work. It would be neat if there was a tool that just asked you a few questions, then took that info and gave you a roadmap, i.e. "Feed Forward Neural Networks, Digit Classification, Image Classification w/ Inception, Object Detection with ResNet + Inception, Optimizing TensorFlow code for Servers, Deploying TensorFlow with Docker, Protecting Against Adversarial Input" This way someone with a time sensitive project doesn't have to learn TF for 6 months before being able to accomplish what they wanted! Just something I think would be neat and also possible to add to TF World.
- madmadjo 8y ago+1 for this one!
- jorgemf 8y agoThere are not so many type of problems for such a complex tool. Your problem usually fits in one category among classification, prediction, clustering, generation or control. Then you have different domains as images, video, audio, text, etc. With a combination or type of problem and domain you sure can have a roadmap, but you probably will need to read papers to solve your problem if it is not something some has done before.
- Aduket 8y agoWhat do you mean by “control”? Like control engineering? How is it used in control domain, can you explain a bit?
- m0zg 8y agoSome well considered advice: drop TensorFlow and go with PyTorch. Spend your effort where it will make a difference: on deep learning, rather than on fighting with the framework. People just keep using TF because it was the first full-fledged Python framework for this, not because it has any technical merit anymore. In PyTorch you will make twice as much progress in half the time.
- jorgemf 8y agoIs there any way to use a pytorch model in Mobile and in a website without a server API? For me 5hose are two good reasons to keep using TensorFlow.
- m0zg 8y agohttps://caffe2.ai/docs/AI-Camera-demo-android.html https://caffe2.ai/docs/AI-Camera-demo-android.html? Not something I've used myself, but supposedly yes.
- jorgemf 8y agoThanks. I always forgot about caffe2 when talking about pytorch. I couldn't anything for JavaScript and mobile seems not as good supported as TF but for sure they will improve.
- m0zg 8y agoHave you actually used TFLite? It's slow as molasses. Deploying on mobile is a bit of a shitshow across the board right now, from what I undertand. Not all models are supported out of the box (especially with ONNX), and the ones that are supported aren't guaranteed to have acceptable performance with off-the-shelf frameworks. Documentation is very sparse as well, especially for the quantized stuff.
- x3tm 8y agoHow do you use TF on mobile? via google.colab?