Abstract: A semiconductor device includes a substrate, an interfacial layer formed on the semiconductor substrate, and a high-k dielectric layer formed on the interfacial layer. At least one of the high-k dielectric layer and the interfacial layer is doped with: a first dopant species, a second dopant species, and a third dopant species. The first dopant species and the second dopant species form a plurality of first dipole elements having a first polarity. The third dopant species forms a plurality of second dipole elements having a second polarity. A first concentration ratio of the first concentration of the first dopant species to the second concentration of the second dopant species of the p-type transistor is different from a second concentration ratio of the first concentration of the first dopant species to the second concentration of the second dopant species of the n-type transistor.

Abstract: Several embodiments include a portable security device. The portable security device can include one or more sensors. The portable security device can compute a home rhythm pattern utilizing a machine learning engine based on a historical record of real-time sensor feeds. The portable security device can camouflage itself as a digital clock, a digital calendar, or a home security dashboard. The portable security device can define an action trigger that binds a state of the environment around the portable security device to at least a device component action. The portable security device can identify a real-time state of the portable security device amongst a finite set of potential states based on features observed from the sensor feeds. The portable security device can execute the device component action at the portable security device in response to determining that the real-time state matches the action trigger.

Abstract: Several embodiments include a portable security device. The portable security device can include one or more sensors. The portable security device can compute a home rhythm pattern utilizing a machine learning engine based on a historical record of real-time sensor feeds. The portable security device can camouflage itself as a digital clock, a digital calendar, or a home security dashboard. The portable security device can define an action trigger that binds a state of the environment around the portable security device to at least a device component action. The portable security device can identify a real-time state of the portable security device amongst a finite set of potential states based on features observed from the sensor feeds. The portable security device can execute the device component action at the portable security device in response to determining that the real-time state matches the action trigger.

Abstract: Several embodiments include a portable security device. The portable security device can include one or more sensors. The portable security device can compute a home rhythm pattern utilizing a machine learning engine based on a historical record of real-time sensor feeds. The portable security device can camouflage itself as a digital clock, a digital calendar, or a home security dashboard. The portable security device can define an action trigger that binds a state of the environment around the portable security device to at least a device component action. The portable security device can identify a real-time state of the portable security device amongst a finite set of potential states based on features observed from the sensor feeds. The portable security device can execute the device component action at the portable security device in response to determining that the real-time state matches the action trigger.

Abstract: Several embodiments include a portable security device. The portable security device can include one or more sensors. The portable security device can compute a home rhythm pattern utilizing a machine learning engine based on a historical record of real-time sensor feeds. The portable security device can camouflage itself as a digital clock, a digital calendar, or a home security dashboard. The portable security device can define an action trigger that binds a state of the environment around the portable security device to at least a device component action. The portable security device can identify a real-time state of the portable security device amongst a finite set of potential states based on features observed from the sensor feeds. The portable security device can execute the device component action at the portable security device in response to determining that the real-time state matches the action trigger.

Abstract: A camera module includes a substrate, a holder, and a dustproof member. The holder is mounted on the substrate. The holder also defines an air vent. The dustproof member includes a permeable portion. The permeable portion is made of air-permeable material. The permeable portion shields the air vent. The dustproof member can also be made of a porous material plugging the air vent.

Abstract: A camera module includes a substrate, a holder, and a dustproof member. The holder is mounted on the substrate. The holder also defines an air vent. The dustproof member includes a permeable portion. The permeable portion is made of air-permeable material. The permeable portion shields the air vent. The dustproof member can also be made of a porous material plugging the air vent.