Abstract: A mechanism is described for facilitating slimming of neural networks in machine learning environments. A method of embodiments, as described herein, includes learning a first neural network associated with machine learning processes to be performed by a processor of a computing device, where learning includes analyzing a plurality of channels associated with one or more layers of the first neural network. The method may further include computing a plurality of scaling factors to be associated with the plurality of channels such that each channel is assigned a scaling factor, wherein each scaling factor to indicate relevance of a corresponding channel within the first neural network. The method may further include pruning the first neural network into a second neural network by removing one or more channels of the plurality of channels having low relevance as indicated by one or more scaling factors of the plurality of scaling factors assigned to the one or more channels.

Abstract: One or more implementations of the present specification provide a method and apparatus for securely entering a trusted execution environment in a hyper-threading scenario. The method can include: in response to that a logical processor running on a physical processor core generates a trusted execution environment entry event through an approach provided by a virtual machine monitor, labeling the logical processor with a state of expecting to enter a trusted execution environment; and in response to determining that all logical processors corresponding to the physical processor core are labeled with the state of expecting to enter a trusted execution environment, separately controlling each one of the logical processors to enter a trusted execution environment built on the physical processor core.

Abstract: An electric vehicle computing sharing system (100) is adapted to receive a signal indicating the electric vehicle (110, 120, 130) is connected to a charging station (115, 125, 135). The computing sharing system (100) may be further adapted to receive information about the electric vehicle (110, 120, 130). The computing sharing system (100) may be further adapted to determine a predicted charging duration (535) for the electric vehicle (110, 120, 130). The computing sharing system (100) may be further adapted to identify a task for execution by a computing resource of the electric vehicle (110, 120, 130) based on the predicted charging duration (535). The computing sharing system (100) may be further adapted to transmit the task to the electric vehicle (110, 120, 130). The computing sharing system (100) may be further adapted to receive a result for the task from the electric vehicle (110, 120, 130).

Abstract: Autonomous robots and methods of operating the same are disclosed. An autonomous robot includes a sensor and memory including machine readable instructions. The autonomous robot further includes at least one processor to execute the instructions to generate a velocity costmap associated with an environment in which the robot is located. The processor generates the velocity costmap based on a source image captured by the sensor. The velocity costmap includes velocity information indicative of movement of an obstacle detected in the environment.

Abstract: Method, systems and apparatuses may provide for technology that divides an application into a plurality of portions that are each associated with one or more functions of the application and determine a plurality of transition probabilities between the plurality of portions. Some technology may also receive at least a first portion of the plurality of portions, and receive a relation file indicating the plurality of transition probabilities between the plurality of portions.

Abstract: One or more implementations of the present specification provide a method and apparatus for securely entering a trusted execution environment in a hyper-threading scenario. The method can include: in response to that a logical processor running on a physical processor core generates a trusted execution environment entry event through an approach provided by a virtual machine monitor, labeling the logical processor with a state of expecting to enter a trusted execution environment; and in response to determining that all logical processors corresponding to the physical processor core are labeled with the state of expecting to enter a trusted execution environment, separately controlling each one of the logical processors to enter a trusted execution environment built on the physical processor core.

Abstract: Method, systems and apparatuses may provide for technology that divides an application into a plurality of portions that are each associated with one or more functions of the application and determine a plurality of transition probabilities between the plurality of portions. Some technology may also receive at least a first portion of the plurality of portions, and receive a relation file indicating the plurality of transition probabilities between the plurality of portions.

Abstract: A method for scheduling a computational task is proposed. The method includes receiving, at a server, a request for executing a computational task from a client device. The method further includes forwarding the computational task to a processing device if a predetermined condition is fulfilled. The predetermined condition can be based on an execution time or on a security level of data of the computational task, for example.

Abstract: Technologies for framework-level audio device virtualization include a computing device that executes multiple application framework instances. The computing device monitors for an application framework instance switch and, in response to an application framework instance switch, determines whether the current application framework instance is in the foreground. If in the foreground, the computing device selects a physical audio output device. The computing device may output audio data associated with the current application framework instance using a kernel audio driver associated with the physical audio output device. If not in the foreground, the computing device selects a null audio output device using a null audio hardware abstraction layer (HAL). The null audio HAL may sleep for the duration of audio data associated with the current application framework instance. The null audio HAL may be an operating-system- and device-independent shared library of the computing device.

Abstract: Preemptive scheduling enclaves as disclosed herein support both cooperative and preemptive scheduling of in-enclave (IE) thread execution. These preemptive scheduling enclaves may include a scheduler configured to be executed as part of normal hardware interrupt processing by enclave threads. The scheduler identifies an IE thread to be scheduled and modifies enclave data structures so that when the enclave thread resumes processing after a hardware interrupt, the identified IE thread is executed, rather than the interrupted IE thread.

Abstract: Technologies for framework-level audio device virtualization include a computing device that executes multiple application framework instances. The computing device monitors for an application framework instance switch and, in response to an application framework instance switch, determines whether the current application framework instance is in the foreground. If in the foreground, the computing device selects a physical audio output device. The computing device may output audio data associated with the current application framework instance using a kernel audio driver associated with the physical audio output device. If not in the foreground, the computing device selects a null audio output device using a null audio hardware abstraction layer (HAL). The null audio HAL may sleep for the duration of audio data associated with the current application framework instance. The null audio HAL may be an operating-system- and device-independent shared library of the computing device.

Abstract: Autonomous robots and methods of operating the same are disclosed. An autonomous robot includes a sensor and memory including machine readable instructions. The autonomous robot further includes at least one processor to execute the instructions to generate a velocity costmap associated with an environment in which the robot is located. The processor generates the velocity costmap based on a source image captured by the sensor. The velocity costmap includes velocity information indicative of movement of an obstacle detected in the environment.

Abstract: A mechanism is described for facilitating slimming of neural networks in machine learning environments. A method of embodiments, as described herein, includes learning a first neural network associated with machine learning processes to be performed by a processor of a computing device, where learning includes analyzing a plurality of channels associated with one or more layers of the first neural network. The method may further include computing a plurality of scaling factors to be associated with the plurality of channels such that each channel is assigned a scaling factor, wherein each scaling factor to indicate relevance of a corresponding channel within the first neural network. The method may further include pruning the first neural network into a second neural network by removing one or more channels of the plurality of channels having low relevance as indicated by one or more scaling factors of the plurality of scaling factors assigned to the one or more channels.

Abstract: Preemptive scheduling enclaves as disclosed herein support both cooperative and preemptive scheduling of in-enclave (IE) thread execution. These preemptive scheduling enclaves may include a scheduler configured to be executed as part of normal hardware interrupt processing by enclave threads. The scheduler identifies an IE thread to be scheduled and modifies enclave data structures so that when the enclave thread resumes processing after a hardware interrupt, the identified IE thread is executed, rather than the interrupted IE thread.

Abstract: Methods, systems, or apparatus may be directed to hosting, by a virtual machine manager of a local machine, a virtual machine having a device driver. A virtual machine manager may obtain, from a stub driver on a remote machine, information about the I/O device on the remote machine. The I/O device may be bound to a stub driver on the remote machine. The virtual machine manager may instantiate a virtual I/O device on the local machine corresponding to the I/O device on the remote machine. The virtual machine manager may then collaborate with the stub driver on the remote machine to effectuate a real access to the I/O device on the remote machine for an access to the virtual I/O device by the device driver on behalf of a program on the local machine.

Abstract: Various embodiments are generally directed to providing virtualization using relatively minimal processing and storage resources to enable concurrent isolated execution of multiple application routines in which one of the application routines is made visible at a time. An apparatus to virtualize an operating system includes a processor component, a visibility checker for execution by the processor component to make a visibility check call to a kernel routine to request an indication of whether an instance of a framework routine that comprises the visibility checker is visible, and resource access code of the instance for execution by the processor component to perform a resource access operation to access a hardware component based on the indication and on receipt of an application programming interface (API) call from an application routine that specifies an API function to access the hardware component. Other embodiments are described and claimed.

Abstract: A method for scheduling a computational task is proposed. The method includes receiving, at a server, a request for executing a computational task from a client device. The method further includes forwarding the computational task to a processing device if a predetermined condition is fulfilled. The predetermined condition can be based on an execution time or on a security level of data of the computational task, for example.

Abstract: Technologies for framework-level audio device virtualization include a computing device that executes multiple application framework instances. The computing device monitors for an application framework instance switch and, in response to an application framework instance switch, determines whether the current application framework instance is in the foreground. If in the foreground, the computing device selects a physical audio output device. The computing device may output audio data associated with the current application framework instance using a kernel audio driver associated with the physical audio output device. If not in the foreground, the computing device selects a null audio output device using a null audio hardware abstraction layer (HAL). The null audio HAL may sleep for the duration of audio data associated with the current application framework instance. The null audio HAL may be an operating-system- and device-independent shared library of the computing device.

Abstract: Various systems and methods for providing access control are described herein. A system comprises a display; a processor; and a memory, including instructions, which when executed on the processor, cause the processor to: present a limited lock screen on a display of the user device, wherein the limited lock screen only provides a non-personalized access mechanism; receive user input via the limited lock screen; correlate the user input with an operating context, wherein the user input is uniquely correlated with the operating context; and unlock the user device with access to the operating context.

Abstract: A computer system may comprise a computer platform and input-output devices. The computer platform may include a plurality of heterogeneous processors comprising a central processing unit (CPU) and a graphics processing unit (GPU) and a shared virtual memory supported by a physical private memory space of at least one heterogeneous processor or a physical shared memory shared by the heterogeneous processor. The CPU (producer) may create shared multi-version data and store such shared multi-version data in the physical private memory space or the physical shared memory. The GPU (consumer) may acquire or access the shared multi-version data.