Questions and Answers
Hayden Liu If you’re still new to ML and DL (like myself), Is it better to know TensorFlow and Pytorch or to focus on just one? And which one would you say is easier to implement?
Glad you ask Ramsi. There is no definite answer to this. However, for ML or DL beginners, PyTorch has an easier learning curve as it is more Pythonic and debug-friendly and more straightforward when it comes to handling data. On the other hand, TensorFlow could indicate a steeper learning curve because of the low-level implementations of the neural network structure. Although TensorFlow has the high-level Keras API, which makes it easy to get started learning the basic concepts, PyTorch has a good trade-off between being easy to use and being more easily customizable than Keras.
So picking either one is fine, don’t need to focus both
Thanks Hayden Liu! I had started with Pytorch some months back, did a couple of projects and moved onto TensorFlow (my internship projects were all in TensorFlow), I’ve been thinking if I need to go back and improve my skills in Pytorch , glad to hear that juggling both early on is not a requirement!
Thanks again :)
Hayden Liu thanks for presenting your book! When it comes to ML, for me at least, the more practical the book the better. Most of the people I know started with “Hands on Machine Learning …” from A.Geron which is also very practical. Could you compare your book to this one? Maybe some pros/cons or why people should read one or the other? Or both?
Hayden Liu? 🙂
Hi Hayden Liu. Given the practical focus of your book, and as someone with very little theoretical knowledge of ML, would you recommend complementing this book with a more theoretical intro book? I suppose I’m worried about falling into the trap of knowing how to implement certain ML algorithms and not the ‘why’ behind it all.
I think it’s a tricky equation balancing the need to ‘get your hands dirty’ and move quickly, with also having solid fundamental (mathematical?) knowledge to underpin it. I was interested in hearing your thoughts on this as I’m someone with plenty of Python engineering experience but very new to ML. Thank you :)
Hi James. Thanks for your question. A lot of readers wonder the same. You can dive into the book with little theoretical knowledge of ml. As each chapter of the book explains the mechanism ML theory and models at the beginning, with the help of examples and codes (instead of pure math), followed by implementations (from scratch and/or with ML packages) and applications. As long as you know Python, and follow the recommended reading list (basic stats and probability) in chapter 1, you are good to go.
Hello Hayden Liu, thanks for being here. You are presenting industry-adopted examples in your book. Do you also talk about how these models are deployed/served and used by clients?
Thanks Philip for the question. There are sections on setting up / running model training and inferencing in GPU, data processing, model training/inferencing on Spark.
Are you referring to serving in API (such as Flask, REST)? Due to the page limit, we decided to keep the book focused on ml thoery and implementations, and applications. But I think you could find tons of one-pager resources online, and super easy to follow
Hello Hayden Liu, thanks for coming to answer our questions about your book. Is there an order for the project examples on your book that you recommend for a ML beginner?
Hey. thanks for asking. It’s a great question btw. You can simply follow the chapters. as they are in the order of difficulty and complexity.
Hi Hayden Liu , thanks for taking the time to answer questions here! Sounds like a great book. Does it also touch on issues related to deploying models in production?
Thanks Allan for the question. There are sections on setting up / running model training and inferencing in GPU, data processing, model training/inferencing on Spark.
Regarding deploying production, due to the page limit, we decided to keep the book focused on ml thoery and implementations, and applications. And in fact, it is rare that an ML book covers it, as it is really stack specific, environment specific. But I think you could find tons of one-pager resources online, and super easy to follow.