I’m going through a similar process. Over the years, I had fits and starts trying to relearn what I had forgotten and clarify what I didn’t intuitively get. With the usual caveats about different people having different learning paths, I didn’t find the knowledge stuck with me if I tried picking a broad math topic (ex: “statistics”, which encompasses a lot) and trying to re-learn everything about it.
Over the last few months, I realized a different approach that had been most effective for me. If I picked a specific thing I found challenging to do or wanted to develop as a skill, I could work backward from there to probe through ever more fundamental concepts where I had a gap in knowledge until I hit something I didn’t need explained to me. Having gone through that backward journey, the path forward now was more clear and obvious. It helped as well that I was motivated to dig deeper because increasing my knowledge within this narrow context had an immediate practical utility for me.
If you would like to take this approach, you need to start somewhere specific. For example, I had a generic goal to learn how I could deploy machine learning in an actually valuable way. After many failed attempts to meet that goal satisfactorily, I decided to pop open MindNode, write TensorFlow 2.0 in the main node, pull up the TF 2.0 Alpha documentation, and start reading from the top. For each thing I didn’t immediately understand (ex: eager execution, input layer shape), I created a node. For each node, I began Googling and reading pretty much everything on the first page of results. If I encountered something within the explanations I didn’t get, that became a new child node. Among the many things I explored, I realized a grain that stuck in my mind was “but why do the rules for matrix multiplication feel so arbitrary”. Exploring that question led me to 3Blue1Brown’s linear algebra playlist on YouTube. I can’t begin to describe how it felt when those videos helped me “get” matrix multiplication. With that fundamental bit of math starting to make sense, the more complex concepts are becoming clearer too.
In a nutshell, the approach I’m taking is to start with something specific that’s personally meaningful, and then dig in to it to find out what exactly isn’t clear to me. I hope it helps to think about this approach as an option.
Over the last few months, I realized a different approach that had been most effective for me. If I picked a specific thing I found challenging to do or wanted to develop as a skill, I could work backward from there to probe through ever more fundamental concepts where I had a gap in knowledge until I hit something I didn’t need explained to me. Having gone through that backward journey, the path forward now was more clear and obvious. It helped as well that I was motivated to dig deeper because increasing my knowledge within this narrow context had an immediate practical utility for me.
If you would like to take this approach, you need to start somewhere specific. For example, I had a generic goal to learn how I could deploy machine learning in an actually valuable way. After many failed attempts to meet that goal satisfactorily, I decided to pop open MindNode, write TensorFlow 2.0 in the main node, pull up the TF 2.0 Alpha documentation, and start reading from the top. For each thing I didn’t immediately understand (ex: eager execution, input layer shape), I created a node. For each node, I began Googling and reading pretty much everything on the first page of results. If I encountered something within the explanations I didn’t get, that became a new child node. Among the many things I explored, I realized a grain that stuck in my mind was “but why do the rules for matrix multiplication feel so arbitrary”. Exploring that question led me to 3Blue1Brown’s linear algebra playlist on YouTube. I can’t begin to describe how it felt when those videos helped me “get” matrix multiplication. With that fundamental bit of math starting to make sense, the more complex concepts are becoming clearer too.
In a nutshell, the approach I’m taking is to start with something specific that’s personally meaningful, and then dig in to it to find out what exactly isn’t clear to me. I hope it helps to think about this approach as an option.