Research
Papers, benchmarks and the techniques behind what eventually ships.
-
The support vector machine is published
Cortes and Vapnik describe the support-vector network for classification.
-
MNIST becomes the field’s default benchmark
LeCun and colleagues release the MNIST digit dataset alongside LeNet-5 results.
-
IBM makes translation a statistical problem
Brown and colleagues publish a statistical, corpus-based approach to machine translation.
-
The Boltzmann machine trains hidden layers
Ackley, Hinton and Sejnowski publish a learning algorithm for stochastic multilayer networks.
-
Hopfield networks revive neural network research
John Hopfield describes a recurrent network storing memories as stable energy states.
-
Shannon shows how a computer could play chess
Claude Shannon publishes search and evaluation strategies for computer chess.
-
McCarthy invents Lisp
John McCarthy designs Lisp, a language built on lists that treats code as data.
-
McCulloch and Pitts model the artificial neuron
McCulloch and Pitts publish a logical model of neurons as binary threshold units.
-
Hebb proposes that neurons learn by firing together
Hebb’s Organization of Behavior proposes co-active neurons strengthen their connections.
-
Injected plans can slip past chain-of-thought monitors, researchers report
A new preprint reports 25 to 33 per cent monitor evasion by hiding a harmful plan in a model’s context, and finds…