Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating a final classification output for an image of eye tissue. The image is provided as input to each of one or more segmentation neural networks to obtain one or more segmentation maps of the eye tissue in the image. A respective classification input is generated from each of the segmentation maps. For each of the segmentation maps, the classification input for the segmentation map is provided as input to each of one or more classification neural networks to obtain, for each segmentation map, a respective classification output from each classification neural network. A final classification output for the image is generated from the respective classification outputs for each of the segmentation maps.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for augmenting neural networks with an external memory. One of the methods includes providing an output derived from the neural network output for the time step as a system output for the time step; maintaining a current state of the external memory; determining, from the neural network output for the time step, memory state parameters for the time step; updating the current state of the external memory using the memory state parameters for the time step; reading data from the external memory in accordance with the updated state of the external memory; and combining the data read from the external memory with a system input for the next time step to generate the neural network input for the next time step.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for augmenting neural networks with an external memory. One of the methods includes providing an output derived from the neural network output for the time step as a system output for the time step; maintaining a current state of the external memory; determining, from the neural network output for the time step, memory state parameters for the time step; updating the current state of the external memory using the memory state parameters for the time step; reading data from the external memory in accordance with the updated state of the external memory; and combining the data read from the external memory with a system input for the next time step to generate the neural network input for the next time step.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for distributed training of reinforcement learning systems. One of the methods includes receiving, by a learner, current values of the parameters of the Q network from a parameter server, wherein each learner maintains a respective learner Q network replica and a respective target Q network replica; updating, by the learner, the parameters of the learner Q network replica maintained by the learner using the current values; selecting, by the learner, an experience tuple from a respective replay memory; computing, by the learner, a gradient from the experience tuple using the learner Q network replica maintained by the learner and the target Q network replica maintained by the learner; and providing, by the learner, the computed gradient to the parameter server.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating a final classification output for an image of eye tissue. The image is provided as input to each of one or more segmentation neural networks to obtain one or more segmentation maps of the eye tissue in the image. A respective classification input is generated from each of the segmentation maps. For each of the segmentation maps, the classification input for the segmentation map is provided as input to each of one or more classification neural networks to obtain, for each segmentation map, a respective classification output from each classification neural network. A final classification output for the image is generated from the respective classification outputs for each of the segmentation maps.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for using a machine learning model that has been trained through reinforcement learning to select a content item. One of the methods includes receiving first data characterizing a first context in which a first content item may be presented to a first user in a presentation environment; and providing the first data as input to a long-term engagement machine learning model, the model having been trained through reinforcement learning to: receive a plurality of inputs, and process each of the plurality of inputs to generate a respective engagement score for each input that represents a predicted, time-adjusted total number of selections by the respective user of future content items presented to the respective user in the presentation environment if the respective content item is presented in the respective context.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating a final classification output for an image of eye tissue. The image is provided as input to each of one or more segmentation neural networks to obtain one or more segmentation maps of the eye tissue in the image. A respective classification input is generated from each of the segmentation maps. For each of the segmentation maps, the classification input for the segmentation map is provided as input to each of one or more classification neural networks to obtain, for each segmentation map, a respective classification output from each classification neural network. A final classification output for the image is generated from the respective classification outputs for each of the segmentation maps.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for using a machine learning model that has been trained through reinforcement learning to select a content item. One of the methods includes receiving first data characterizing a first context in which a first content item may be presented to a first user in a presentation environment; and providing the first data as input to a long-term engagement machine learning model, the model having been trained through reinforcement learning to: receive a plurality of inputs, and process each of the plurality of inputs to generate a respective engagement score for each input that represents a predicted, time-adjusted total number of selections by the respective user of future content items presented to the respective user in the presentation environment if the respective content item is presented in the respective context.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for selecting answers to questions about documents. One of the methods includes receiving a document comprising a plurality of document tokens; receiving a question associated with the document, the question comprising a plurality of question tokens; processing the document tokens and the question tokens using a reader neural network to generate a joint numeric representation of the document and the question; and selecting, from the plurality of document tokens, an answer to the question using the joint numeric representation of the document and the question.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for distributed training of reinforcement learning systems. One of the methods includes receiving, by a learner, current values of the parameters of the Q network from a parameter server, wherein each learner maintains a respective learner Q network replica and a respective target Q network replica; updating, by the learner, the parameters of the learner Q network replica maintained by the learner using the current values; selecting, by the learner, an experience tuple from a respective replay memory; computing, by the learner, a gradient from the experience tuple using the learner Q network replica maintained by the learner and the target Q network replica maintained by the learner; and providing, by the learner, the computed gradient to the parameter server.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for augmenting neural networks with an external memory. One of the methods includes providing an output derived from the neural network output for the time step as a system output for the time step; maintaining a current state of the external memory; determining, from the neural network output for the time step, memory state parameters for the time step; updating the current state of the external memory using the memory state parameters for the time step; reading data from the external memory in accordance with the updated state of the external memory; and combining the data read from the external memory with a system input for the next time step to generate the neural network input for the next time step.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for distributed training of reinforcement learning systems. One of the methods includes receiving, by a learner, current values of the parameters of the Q network from a parameter server, wherein each learner maintains a respective learner Q network replica and a respective target Q network replica; updating, by the learner, the parameters of the learner Q network replica maintained by the learner using the current values; selecting, by the learner, an experience tuple from a respective replay memory; computing, by the learner, a gradient from the experience tuple using the learner Q network replica maintained by the learner and the target Q network replica maintained by the learner; and providing, by the learner, the computed gradient to the parameter server.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for augmenting neural networks with an external memory. One of the methods includes providing an output derived from the neural network output for the time step as a system output for the time step; maintaining a current state of the external memory; determining, from the neural network output for the time step, memory state parameters for the time step; updating the current state of the external memory using the memory state parameters for the time step; reading data from the external memory in accordance with the updated state of the external memory; and combining the data read from the external memory with a system input for the next time step to generate the neural network input for the next time step.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating a final classification output for an image of eye tissue. The image is provided as input to each of one or more segmentation neural networks to obtain one or more segmentation maps of the eye tissue in the image. A respective classification input is generated from each of the segmentation maps. For each of the segmentation maps, the classification input for the segmentation map is provided as input to each of one or more classification neural networks to obtain, for each segmentation map, a respective classification output from each classification neural network. A final classification output for the image is generated from the respective classification outputs for each of the segmentation maps.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating a final classification output for an image of eye tissue. The image is provided as input to each of one or more segmentation neural networks to obtain one or more segmentation maps of the eye tissue in the image. A respective classification input is generated from each of the segmentation maps. For each of the segmentation maps, the classification input for the segmentation map is provided as input to each of one or more classification neural networks to obtain, for each segmentation map, a respective classification output from each classification neural network. A final classification output for the image is generated from the respective classification outputs for each of the segmentation maps.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating a final classification output for an image of eye tissue. The image is provided as input to each of one or more segmentation neural networks to obtain one or more segmentation maps of the eye tissue in the image. A respective classification input is generated from each of the segmentation maps. For each of the segmentation maps, the classification input for the segmentation map is provided as input to each of one or more classification neural networks to obtain, for each segmentation map, a respective classification output from each classification neural network. A final classification output for the image is generated from the respective classification outputs for each of the segmentation maps.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for using a machine learning model that has been trained through reinforcement learning to select a content item. One of the methods includes receiving first data characterizing a first context in which a first content item may be presented to a first user in a presentation environment; and providing the first data as input to a long-term engagement machine learning model, the model having been trained through reinforcement learning to: receive a plurality of inputs, and process each of the plurality of inputs to generate a respective engagement score for each input that represents a predicted, time-adjusted total number of selections by the respective user of future content items presented to the respective user in the presentation environment if the respective content item is presented in the respective context.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for selecting answers to questions about documents. One of the methods includes receiving a document comprising a plurality of document tokens; receiving a question associated with the document, the question comprising a plurality of question tokens; processing the document tokens and the question tokens using a reader neural network to generate a joint numeric representation of the document and the question; and selecting, from the plurality of document tokens, an answer to the question using the joint numeric representation of the document and the question.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for augmenting neural networks with an external memory. One of the methods includes providing an output derived from the neural network output for the time step as a system output for the time step; maintaining a current state of the external memory; determining, from the neural network output for the time step, memory state parameters for the time step; updating the current state of the external memory using the memory state parameters for the time step; reading data from the external memory in accordance with the updated state of the external memory; and combining the data read from the external memory with a system input for the next time step to generate the neural network input for the next time step.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for distributed training of reinforcement learning systems. One of the methods includes receiving, by a learner, current values of the parameters of the Q network from a parameter server, wherein each learner maintains a respective learner Q network replica and a respective target Q network replica; updating, by the learner, the parameters of the learner Q network replica maintained by the learner using the current values; selecting, by the learner, an experience tuple from a respective replay memory; computing, by the learner, a gradient from the experience tuple using the learner Q network replica maintained by the learner and the target Q network replica maintained by the learner; and providing, by the learner, the computed gradient to the parameter server.