I've spent the past 2 days writing code to work with OCR results and use the position of words and sentences to link them together and classify them.<p>In the process my math knowledge has had a real workout, even though it's 2D geometry.<p>I did undergrad physics at university 20 years ago. Since then I've done web programming that hasn't needed maths and I've forgotten so much. Though I don't feel I ever understood the maths properly, just enough to pass my courses.<p>Can you recommend any books or learning materials to help me get back in practice? I'm a visual learner and learn best with applied examples, which is where many of my textbooks fell down as they were too abstract. Anything that's based on doing maths on images or something I can see and draw results on is even better.<p>I'm also looking for material that covers how you apply maths on computers. E.g. I was extrapolating a line through two points and got caught out when the gradient went to infinity (in x-y coords, a line parallel to the y-axis). I fudged it by adding a tiny amount to the denominator so it'll never by zero, but I'd like to learn the proper way to handle these cases.
Translating problems into vector representations would be good too, as I tried and got lost in the different meaning of Vector between physics and math.
Maybe check out the book:<p>Coding the Matrix: Linear Algebra through Applications to Computer Science<p><a href="https://codingthematrix.com/" rel="nofollow">https://codingthematrix.com/</a>
Look like reviewing a bit of linear algebra would be helpful to you. The extrapolation problem is one of that. Assuming (x1,y1) and (x2,y2), the line you are looking for is (x,y) = a * (x1,y1) + (1-a) * (x2,y2) is a LA problem. There are many books on LA out there, i learned from Gilbert Strang, but not sure how deep you’d have to go.