Abstract: A method for residual adapters for few-shot text-to-speech speaker adaptation includes obtaining a text-to-speech (TTS) model configured to convert text into representations of synthetic speech, the TTS model pre-trained on an initial training data set. The method further includes augmenting the TTS model with a stack of residual adapters. The method includes receiving an adaption training data set including one or more spoken utterances spoken by a target speaker, each spoken utterance in the adaptation training data set paired with corresponding input text associated with a transcription of the spoken utterance. The method also includes adapting, using the adaption training data set, the TTS model augmented with the stack of residual adapters to learn how to synthesize speech in a voice of the target speaker by optimizing the stack of residual adapters while parameters of the TTS model are frozen.

Abstract: A method for residual adapters for few-shot text-to-speech speaker adaptation includes obtaining a text-to-speech (TTS) model configured to convert text into representations of synthetic speech, the TTS model pre-trained on an initial training data set. The method further includes augmenting the TTS model with a stack of residual adapters. The method includes receiving an adaption training data set including one or more spoken utterances spoken by a target speaker, each spoken utterance in the adaptation training data set paired with corresponding input text associated with a transcription of the spoken utterance. The method also includes adapting, using the adaption training data set, the TTS model augmented with the stack of residual adapters to learn how to synthesize speech in a voice of the target speaker by optimizing the stack of residual adapters while parameters of the TTS model are frozen.

Abstract: Methods, systems, and computer-readable media for optimizing diversity and relevance for catalog items are disclosed. Respective target metrics are determined for a plurality of categories comprising a plurality of items in an electronic catalog. First and second categories comprise first and second subsets of the plurality of items, and the respective target metric for the first category is determined using aggregation of individual target metrics for the first subset. Respective similarity scores are determined between pairs of the categories. N item categories are selected from the plurality of categories based at least in part on the respective target metrics and based at least in part on the respective similarity scores. A third subset of the plurality of items is selected from across the N item categories, including the first and second categories. The selected items optimize both category diversity and item relevance.

Abstract: Provided is an evaluation apparatus comprising: an environmental data acquisition unit configured to acquire environmental data indicating an operation environment of equipment; a performance data acquisition unit configured to acquire performance data indicating an operation performance of the equipment; an estimation unit configured to estimate an operation performance on an operation basis in a learning target period under an operation environment in an evaluation target period, based on the environmental data and the performance data in the learning target period; an evaluation unit configured to calculate an indicator that relatively evaluates an actually measured value of an operation performance in the evaluation target period with respect to an estimated value of the estimated operation performance; and an output unit configured to output the indicator.

Abstract: The techniques described herein determine an initial set of ranked images associated with an image-based search query. Based on visual content similarities between images in the initial set of ranked images, the techniques select confident image samples from the initial set of ranked images. The techniques then use the confident image samples to rerank the initial set of ranked images. Accordingly, a search engine uses the confident image samples to promote images that are likely to be relevant to the search query, while demoting images that are not likely to be relevant to the search query. Therefore, the search engine can provide improved relevance-based search results to an image-based search query.

Abstract: An object of the present invention is to provide a deployment shelter which can be changed between a stored state and a deployed state by smoothly driving movable panels. The deployment shelter 10 has a shelter main body 20 in which a floor panel 28 is coupled to a roof panel 29 via four poles. Movable roof panels 37a and 37b which are pivoted between stored positions and deployed positions are respectively attached to side portions of the roof panel 29, movable floor panels 41a and 41b which are pivoted between stored positions and deployed positions are respectively attached to side portions of the floor panel 28. Movable side panels 42a and 42b are respectively attached to the movable floor panels 41a and 41b. Movable end panels 43a and 43b which are pivoted between stored positions and deployed positions are respectively attached to poles. The movable roof panels, the movable floor panels, and the movable side panels are respectively driven by actuators so that they are opened and closed.

Abstract: The techniques described herein determine an initial set of ranked images associated with an image-based search query. Based on visual content similarities between images in the initial set of ranked images, the techniques select confident image samples from the initial set of ranked images. The techniques then use the confident image samples to rerank the initial set of ranked images. Accordingly, a search engine uses the confident image samples to promote images that are likely to be relevant to the search query, while demoting images that are not likely to be relevant to the search query. Therefore, the search engine can provide improved relevance-based search results to an image-based search query.

Abstract: A deployment shelter has: a top panel; and a floor panel provided below the top panel so as to form an internal space between the top panel and the floor panel, the deployment shelter further having: an upper and a lower frame to which the top panel and the floor panel are fixed; a louver board pivotably provided to an end portion of the top panel, the louver board being movable between a stored state and a deployed state; a movable floor panel pivotably provided to an end portion of the floor panel, the movable floor panel being movable between a stored state and a deployed state; a side panel pivotably provided to a free end of the movable floor panel, and movable between a stored state and a deployed state; and a movable wall pivotably provided to the frame, and movable between a stored state and a deployed state.

Abstract: An object of the present invention is to provide a deployment shelter which can be changed between a stored state and a deployed state by smoothly driving movable panels. The deployment shelter 10 has a shelter main body 20 in which a floor panel 28 is coupled to a roof panel 29 via four poles. Movable roof panels 37a and 37b which are pivoted between stored positions and deployed positions are respectively attached to side portions of the roof panel 29, movable floor panels 41a and 41b which are pivoted between stored positions and deployed positions are respectively attached to side portions of the floor panel 28. Movable side panels 42a and 42b are respectively attached to the movable floor panels 41a and 41b. Movable end panels 43a and 43b which are pivoted between stored positions and deployed positions are respectively attached to poles. The movable roof panels, the movable floor panels, and the movable side panels are respectively driven by actuators so that they are opened and closed.

Abstract: A method of controlling traffic signals at a road intersection, which has a plurality of signal groups, each of which controls at least one direction of traffic within the intersection. The method comprises the steps of obtaining and utilizing traffic data to calculate a current traffic state and the rate of change in the traffic state. The method further comprises formulating at least one action and the duration of the action in response to these calculations. Each action comprises switching at least one traffic signal. One or more policies based on the calculations and the action are resolved. A continuous decision making process is applied to evaluate a reward for the policies resolved and a policy that maximizes the reward is selected.

Abstract: A method of controlling traffic signals at a road intersection, which has a plurality of signal groups, each of which controls at least one direction of traffic within the intersection. The method comprises the steps of obtaining and utilising traffic data to calculate a current traffic state and the rate of change in the traffic state. The method further comprises formulating at least one action and the duration of the action in response to these calculations. Each action comprises switching at least one traffic signal. One or more policies based on the calculations and the action are resolved. A continuous decision making process is applied to evaluate a reward for the policies resolved and a policy that maximizes the reward is selected.