Abstract: Generally, the present disclosure is directed to systems and methods that provide a simple, scalable, yet effective strategy to perform transfer learning with a mixture of experts (MoE). In particular, the transfer of pre-trained representations can improve sample efficiency and reduce computational requirements for new tasks. However, representations used for transfer are usually generic, and are not tailored to a particular distribution of downstream tasks. In contrast, example systems and methods of the present disclosure use expert representations for transfer with a simple, yet effective, strategy.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for performing a machine learning task on a network input to generate a network output. In one aspect, one of the systems includes a neural network configured to perform the machine learning task, the neural network including one or more expert neural network blocks that each include multiple routers and multiple expert neural networks.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training neural networks using transfer learning.

Abstract: Techniques are disclosed herein for making predictions with respect to how content consumers will interact with a digital asset. For example, in the context of website visitors browsing digital assets provided via a website, web traffic data can be collected and modeled using a belief network. The belief network may represent a probability distribution for a set of variables that define the web traffic data. Examples of such variables include browser type, browsing session duration, geographic location, visitor demographic characteristics, and a browsing outcome. Certain of the embodiments disclosed herein can be used to extract knowledge from the belief network, thereby allowing statistical inferences to be drawn with respect to how certain classes of website visitors will interact with the website. The influence of one or more first variables (for example, geographic location) can be quantified with respect to one or more second variables (for example, the successful result indicator).

Abstract: A smart office desk interactive with the user, defined on the basis of an individual work desk (M) and various hardware and software elements applied to it and whose main hardware components comprise a working desktop divided into two zones, one which can be raised at an angle, motor-driven legs which raise and lower the desktop, a footrest tray and a series of sensors integrated into the desktop structure and in the legs. The smart office desk learns from the user by machine learning algorithms, and proposes physical actions in accordance with what has been learned. The table collects information from the user concerning their postures and working habits, using various physical sensors, and by their interactions with the software application displayed onscreen. It is thus a table controlled by a computer system which receives data on the user and the working environment as inputs.

Abstract: Techniques are disclosed herein for making predictions with respect to how content consumers will interact with a digital asset. For example, in the context of website visitors browsing digital assets provided via a website, web traffic data can be collected and modeled using a belief network. The belief network may represent a probability distribution for a set of variables that define the web traffic data. Examples of such variables include browser type, browsing session duration, geographic location, visitor demographic characteristics, and a browsing outcome. Certain of the embodiments disclosed herein can be used to extract knowledge from the belief network, thereby allowing statistical inferences to be drawn with respect to how certain classes of website visitors will interact with the website. The influence of one or more first variables (for example, geographic location) can be quantified with respect to one or more second variables (for example, the successful result indicator).

Abstract: Techniques are disclosed herein for making predictions with respect to how content consumers will interact with a digital asset. For example, in the context of website visitors browsing digital assets provided via a website, web traffic data can be collected and modeled using a belief network. The belief network may represent a probability distribution for a set of variables that define the web traffic data. Examples of such variables include browser type, browsing session duration, geographic location, visitor demographic characteristics, and a browsing outcome. Certain of the embodiments disclosed herein can be used to extract knowledge from the belief network, thereby allowing statistical inferences to be drawn with respect to how certain classes of website visitors will interact with the website. The influence of one or more first variables (for example, geographic location) can be quantified with respect to one or more second variables (for example, the successful result indicator).

Abstract: A smart office desk interactive with the user, defined on the basis of an individual work desk (M) and various hardware and software elements applied to it and whose main hardware components comprise a working desktop divided into two zones, one which can be raised at an angle, motor-driven legs which raise and lower the desktop, a footrest tray and a series of sensors integrated into the desktop structure and in the legs. The smart office desk learns from the user by machine learning algorithms, and proposes physical actions in accordance with what has been learned. The table collects information from the user concerning their postures and working habits, using various physical sensors, and by their interactions with the software application displayed onscreen. It is thus a table controlled by a computer system which receives data on the user and the working environment as inputs.

Abstract: Ergonomic desk and modular design for collaborative dispositions, where said ergonomic desk has an upper desktop (1) of trapezoid shape, with an ergonomic curve adapted to the user, a height-adjustable plane (2) in an angle in the middle of said desktop raised by a telescopic linear actuator, a footrest tray (3), a raised tray (6) and motor-driven height-adjustable legs (9) moored to a flat support (8) with wheels (10) and an electronic device with a touch-screen (5) to operate the assembly, characterised by the fact that with a modular coupling of six ergonomic desks, creating a hexagonal unit, a working area of 7.8 m2 is obtained, within a 9.9 m2 square, representing a 38% reduction of the working area created with traditional desks.

Abstract: Techniques are disclosed herein for making predictions with respect to how content consumers will interact with a digital asset. For example, in the context of website visitors browsing digital assets provided via a website, web traffic data can be collected and modeled using a belief network. The belief network may represent a probability distribution for a set of variables that define the web traffic data. Examples of such variables include browser type, browsing session duration, geographic location, visitor demographic characteristics, and a browsing outcome. Certain of the embodiments disclosed herein can be used to extract knowledge from the belief network, thereby allowing statistical inferences to be drawn with respect to how certain classes of website visitors will interact with the website. The influence of one or more first variables (for example, geographic location) can be quantified with respect to one or more second variables (for example, the successful result indicator).