CG: n
CT: These three interwoven processes—a loss function that tracks model error across different inputs, the backward propagation of that error to see how different parts of the network contribute to the error and the gradient descent algorithms that adjust model weights accordingly—are how deep learning models “learn.” As such, backpropagation is fundamental to training neural network models, from the most basic multilayer perceptrons to the complex deep neural network architectures used for generative AI.
S: IBM – https://www.ibm.com/think/topics/backpropagation (last access: 24 January 2025)
N: 1. From “back” (from Old English bæc “back,” from Proto-Germanic *bakam, cognates: Old Saxon and Middle Dutch bak, Old Frisian bek, with no known connections outside Germanic) + “propagation” (from mid-15c., propagacioun, “the causing of plants or animals to reproduce; reproduction; act or fact of begetting or being begotten,” from Old French propagacion “offshoot, offspring” (13c.) and directly from Latin propagationem, nominative propagatio).
- It is shorthand for “backwards propagation of errors”.
2. The term backpropagation refers to a specific algorithm for how weights in a neural network are updated in reverse order at the end of each training epoch. It is a two-step process: propagation and weight update. During propagation, the input data is passed forward through the network, and the output is compared with the expected result. This produces an error value for each neuron in the output layer. During the weight update, this error value is distributed back through the network, giving each neuron an error value that reflects its contribution to the original output. This error value is then used to adjust the weights of each neuron during the next training epoch.
3. Backpropagation is a cornerstone of modern deep learning because it enables efficient training of deep neural networks. By computing gradients through automatic differentiation, it eliminates the need for manual calculation of derivatives, making it feasible to train networks with millions or even billions of parameters.
4. Backpropagation is key for image and pattern recognition, and deep learning models that are trained using it play a significant role in various tasks in self-driving cars (such as object detection and lane detection) or in medical image analysis.
5. Artificial Intelligence : back propagation, back-propagation, feedback propagation, backward propagation.
- In a multilayered network, the propagation of connection weights adjustment, layer by layer, from the output layer toward the input of the network.
- Back-propagation uses simple calculus to decide how to change the parameters of the neural network. It takes a training example – an input and its corresponding output – and makes small modifications to the network parameters to minimize the difference between the current response of the network and the target response.
- Typically back propagation is used in connection with error-correction learning.
- back propagation; feedback propagation: terms and definition standardized by ISO/IEC [International Electrotechnical Commission] in 1999.
S: 1. Etymonline – https://www.etymonline.com/search?q=back, https://www.etymonline.com/search?q=propagation (last access: 24 January 2025); IBM – https://www.ibm.com/think/topics/backpropagation (last access: 24 January 2025). 2. VVG – https://www.vationventures.com/glossary/backpropagation-definition-explanation-and-use-cases (last access: 24 January 2025). 3 & 4. Ultralytics – https://www.ultralytics.com/glossary/backpropagation (last access: 24 January 2025). 5. TERMIUM PLUS – https://www.btb.termiumplus.gc.ca/tpv2alpha/alpha-eng.html?lang=eng&i=1&srchtxt=RETROPROPAGATION&index=alt&codom2nd_wet=1#resultrecs (last access: 24 January 2025).
OV: 1. back-propagation. 2. back propagation.
S: 1. TERMIUM PLUS – https://www.btb.termiumplus.gc.ca/tpv2alpha/alpha-eng.html?lang=eng&i=1&srchtxt=RETROPROPAGATION&index=alt&codom2nd_wet=1#resultrecs (last access: 24 January 2025). 2. GDT – https://vitrinelinguistique.oqlf.gouv.qc.ca/fiche-gdt/fiche/8393872/retropropagation-derreurs (last access: 24 January 2025).
SYN: 1. feedback propagation. 2. backward propagation. 3. error back propagation, error feedback propagation. (depending on context)
S: 1. TERMIUM PLUS – https://www.btb.termiumplus.gc.ca/tpv2alpha/alpha-eng.html?lang=eng&i=1&srchtxt=RETROPROPAGATION&index=alt&codom2nd_wet=1#resultrecs (last access: 24 January 2025); GDT – https://vitrinelinguistique.oqlf.gouv.qc.ca/fiche-gdt/fiche/8393872/retropropagation-derreurs (last access: 24 January 2025). 2. TERMIUM PLUS – https://www.btb.termiumplus.gc.ca/tpv2alpha/alpha-eng.html?lang=eng&i=1&srchtxt=RETROPROPAGATION&index=alt&codom2nd_wet=1#resultrecs (last access: 24 January 2025). 3. GDT – https://vitrinelinguistique.oqlf.gouv.qc.ca/fiche-gdt/fiche/8393872/retropropagation-derreurs (last access: 24 January 2025).
