Deep Neural Networks for YouTube Recommendations
Citations
412 citations
388 citations
Cites background from "Deep Neural Networks for YouTube Re..."
...In addition, it is important to make recommendations in the real world [14] that respond to scalability and noise in learning data [14]....
[...]
383 citations
Cites background from "Deep Neural Networks for YouTube Re..."
...For example, for recommendation systems (Li et al., 2011; Covington et al., 2016) or health applications (Murphy et al., 2001), deploying a new policy may only be done at a low frequency after extensive testing and evaluation....
[...]
...For example, for recommendation systems (Li et al., 2011; Covington et al., 2016) or health applications (Murphy et al....
[...]
380 citations
Cites background from "Deep Neural Networks for YouTube Re..."
...For example, consider the huge action spaces in recommender systems (Covington et al., 2016), or the number of sensors and actuators to con-...
[...]
...This may be on the order of milliseconds for a recommender system (Covington et al., 2016) responding to a user request or the control of a physical robot, and up to the order of minutes for building control systems (Evans & Gao)....
[...]
...For example, consider the huge action spaces in recommender systems (Covington et al., 2016), or the number of sensors and actuators to con- trol cooling in a Google data center (Evans & Gao)....
[...]
377 citations
Cites background from "Deep Neural Networks for YouTube Re..."
...From these works, it is proved that CNNs are able to extract more general contextual features from texts and the features have been successfully used to build recommender systems showing less prediction error than that of CF/MF ([6, 8])....
[...]
...As in CF/MF techniques, the latent representations (γu ,γi ) are mapped into the same vector space (RK ) and the ratings can be estimated by the inner product. r̂u,i = γ > u γi (12) The estimation can be considered as a regression problem and all parameters in the two networks (user network and item network) are trained jointly through the backpropagation technique....
[...]
...Another drawback of CF/MF techniques is content-ignorance....
[...]
References
[...]
139,059 citations
"Deep Neural Networks for YouTube Re..." refers background in this paper
...We observe that the most important signals are those that describe a user’s previous interaction with the item itself and other similar items, matching others’ experience in ranking ads [7]....
[...]
30,843 citations
24,012 citations
"Deep Neural Networks for YouTube Re..." refers background in this paper
...A key advantage of using deep neural networks as a generalization of matrix factorization is that arbitrary continuous and categorical features can be easily added to the model....
[...]
17,184 citations
11,343 citations