This lesson is a general overview of overarching concepts in neuroinformatics research, with a particular focus on clinical approaches to defining, measuring, studying, diagnosing, and treating various brain disorders. Also described are the complex, multi-level nature of brain disorders and the data associated with them, from genes and individual cells up to cortical microcircuits and whole-brain network dynamics. Given the heterogeneity of brain disorders and their underlying mechanisms, this lesson lays out a case for multiscale neuroscience data integration.
This lesson gives an in-depth introduction of ethics in the field of artificial intelligence, particularly in the context of its impact on humans and public interest. As the healthcare sector becomes increasingly affected by the implementation of ever stronger AI algorithms, this lecture covers key interests which must be protected going forward, including privacy, consent, human autonomy, inclusiveness, and equity.
This lesson describes a definitional framework for fairness and health equity in the age of the algorithm. While acknowledging the impressive capability of machine learning to positively affect health equity, this talk outlines potential (and actual) pitfalls which come with such powerful tools, ultimately making the case for collaborative, interdisciplinary, and transparent science as a way to operationalize fairness in health equity.
This lesson breaks down the principles of Bayesian inference and how it relates to cognitive processes and functions like learning and perception. It is then explained how cognitive models can be built using Bayesian statistics in order to investigate how our brains interface with their environment.
This lesson corresponds to slides 1-64 in the PDF below.
This lecture covers a lot of post-war developments in the science of the mind, focusing first on the cognitive revolution, and concluding with living machines.
This lecture provides an overview of depression (epidemiology and course of the disorder), clinical presentation, somatic co-morbidity, and treatment options.
This presentation accompanies the paper entitled: An automated pipeline for constructing personalized virtual brains from multimodal neuroimaging data (see link below to download publication).
This lecture focuses on how to get from a scientific question to a model using concrete examples. We will present a 10-step practical guide on how to succeed in modeling. This lecture contains links to 2 tutorials, lecture/tutorial slides, suggested reading list, and 3 recorded Q&A sessions.
This lecture formalizes modeling as a decision process that is constrained by a precise problem statement and specific model goals. We provide real-life examples on how model building is usually less linear than presented in Modeling Practice I.
This lecture focuses on the purpose of model fitting, approaches to model fitting, model fitting for linear models, and how to assess the quality and compare model fits. We will present a 10-step practical guide on how to succeed in modeling.
This lecture summarizes the concepts introduced in Model Fitting I and adds two additional concepts: 1) MLE is a frequentist way of looking at the data and the model, with its own limitations. 2) Side-by-side comparisons of bootstrapping and cross-validation.
This lecture provides an overview of the generalized linear models (GLM) course, originally a part of the Neuromatch Academy (NMA), an interactive online summer school held in 2020. NMA provided participants with experiences spanning from hands-on modeling experience to meta-science interpretation skills across just about everything that could reasonably be included in the label "computational neuroscience".
This lecture further develops the concepts introduced in Machine Learning I. This lecture is part of the Neuromatch Academy (NMA), an interactive online computational neuroscience summer school held in 2020.
This lecture covers the application of dimensionality reduction applied to multi-dimensional neural recordings using brain-computer interfaces with simultaneous spike recordings.
This is a tutorial covering Generalized Linear Models (GLMs), which are a fundamental framework for supervised learning. In this tutorial, the objective is to model a retinal ganglion cell spike train by fitting a temporal receptive field: first with a Linear-Gaussian GLM (also known as ordinary least-squares regression model) and then with a Poisson GLM (aka "Linear-Nonlinear-Poisson" model). The data you will be using was published by Uzzell & Chichilnisky 2004.