"We've designed a distributed system for sharing enormous datasets - for researchers, by researchers. The result is a scalable, secure, and fault-tolerant repository for data, with blazing fast download speeds." But, Zotero can't discover citation data on their dataset pages; I think I'll drop them a line.
Andrew Ng's Coursera MOOC on Machine Learning covers the same material (even has an exercise on recognizing handwritten digits). There's more meat in this chapter, but Ng's introduction is more gradual and hands-on; both are worth a look.
Amazing recount of the creation of T, the first production-grade Scheme implementation. Funny quote: "I fashionably decried premature optimisation in college without really understanding it until I once committed an act of premature opt so horrific that I can now tell when it is going to rain by the twinges I get in the residual scar tissue. Now I understand premature optimisation." (Olin Shivers)