Learning algorithms

A learning algorithm is a mathematical framework or procedure that calculates the best output given a particular set of data. It does this by updating the calculation based on the difference between the actual and desired output. These algorithms are typically concerned with representation and generalization of the input data.

Latest Research and Reviews

News and Comment

  • News and Views |

    Motor learning is composed of explicit ‘strategic’ components and implicit ‘automatic’ components. Miyamoto and colleagues reveal how these components work together during visuomotor adaptation, providing evidence that an implicit component corrects for a noisy explicit process.

    • Olivier Codol
    • , Giacomo Ariani
    •  & Jonathan A. Michaels
    Nature Neuroscience 23, 298-300
  • News and Views |

    Classic theories of reinforcement learning and neuromodulation rely on reward prediction errors. A new machine learning technique relies on neuromodulatory signals that are optimized for specific tasks, which may lead to better AI and better explanations of neuroscience data.

    • Blake A. Richards
  • News and Views |

    Disproportionate reactions to unexpected stimuli and greater attention to perceived threat are cardinal symptoms of post-traumatic stress disorder. Computational psychiatry helps explain how these responses develop and result from abnormalities in learning and prediction during and after traumatic events.

    • Peggy Seriès
    Nature Neuroscience 22, 334-336
  • News and Views |

    What you choose depends on what information your brain considers and what it neglects when computing the value of actions. An early theory used this insight for a computational account of habits versus deliberation. It has ultimately helped uncover how choice in the brain goes beyond such simple dichotomies.

    • Nathaniel D. Daw
    Nature Neuroscience 21, 1497-1499