https://joelmwaka.com Data scientist. Portfolio, technical notes, and occasional writing on machine learning, statistics, and the craft of working with data. en https://joelmwaka.com/blog/reading-papers-without-drowning https://joelmwaka.com/blog/reading-papers-without-drowning Thu, 02 Jul 2026 00:00:00 GMT A three-pass system for keeping up with machine learning research without letting the backlog win. https://joelmwaka.com/learning/signal-processing/bayesian-parameter-estimation https://joelmwaka.com/learning/signal-processing/bayesian-parameter-estimation Tue, 11 Feb 2025 00:00:00 GMT The parameter becomes a random vector with a prior. Bayes' theorem turns one data record into a posterior, and MAP, MMSE, and LMMSE are three ways of summarizing it, collapsing into a single formula when everything is Gaussian. https://joelmwaka.com/learning/deep-learning/geometry-of-gradient-descent https://joelmwaka.com/learning/deep-learning/geometry-of-gradient-descent Tue, 28 Jan 2025 00:00:00 GMT Why the negative gradient is the direction of steepest descent — and what the learning rate is really controlling. https://joelmwaka.com/learning/signal-processing/classical-parameter-estimation https://joelmwaka.com/learning/signal-processing/classical-parameter-estimation Sat, 28 Sep 2024 00:00:00 GMT The classical branch in full, with MVUE as the goal, the Cramér–Rao bound as the yardstick, maximum likelihood as the workhorse, and least squares when a signal model is all you have. https://joelmwaka.com/learning/signal-processing/parameter-estimation https://joelmwaka.com/learning/signal-processing/parameter-estimation Thu, 30 May 2024 00:00:00 GMT What it means to estimate a parameter, and a map of the six workhorse estimators — LS, MVUE, MLE, LMMSE, MMSE, MAP — chosen by asking what you actually know.