Systems and methods for guiding allocation of computational resources in automated perceptual systems
First Claim
1. An analysis system, comprising:
- a plurality of sensors configured to collect perception evidence;
a computer system comprising a perception system that, when executed, consumes computing resources of the computing system, the perception system comprising;
a control component that selectively gathers from the plurality of sensors the perception evidence, the perception evidence being gathered at times selected to limit utilization of the computing resources of the computer system by the perception system; and
an analysis component that utilizes an analysis policy to analyze the perception evidence to determine a context of a user of the computer system,wherein;
the control component uses information from the analysis component regarding at least one feature about which to collect data to select the times at which to gather perception evidence from one or more of the plurality of sensors;
the analysis component generates the information based on a value of the at least one feature for determining the context;
the analysis policy comprises at least one selected from the group consisting of;
a random selection perception policy that randomly selects which features to utilize on a frame by frame basis, a rate-based perception policy that defines observational frequencies and duty cycles for at least one feature or an EVI-based perception policy that determines an expected value of information via a cost-benefit analysis means utilizing at least expected values and cost of analysis values for at least one feature, the cost of analysis values comprising at least one value proportional to an impact to the computing resources employed by the perception system.
4 Assignments
0 Petitions
Accused Products
Abstract
The present invention leverages analysis methods, such as expected value of information techniques, rate-based techniques, and random selection technique, to provide a fusion of low-level streams of input data (e.g., raw data) from multiple sources to facilitate in inferring human-centric notions of context while reducing computational resource burdens. In one instance of the present invention, the method utilizes real-time computations of expected value of information in a greedy, one-step look ahead approach to compute a next best set of observations to make at each step, producing “EVI based-perception.” By utilizing dynamically determined input data, the present invention provides utility-directed information gathering to enable a significant reduction in system resources. Thus, of the possible input combinations, the EVI-based system can automatically determine which sources are required for real-time computation relating to a particular context.
68 Citations
82 Claims
-
1. An analysis system, comprising:
-
a plurality of sensors configured to collect perception evidence; a computer system comprising a perception system that, when executed, consumes computing resources of the computing system, the perception system comprising; a control component that selectively gathers from the plurality of sensors the perception evidence, the perception evidence being gathered at times selected to limit utilization of the computing resources of the computer system by the perception system; and an analysis component that utilizes an analysis policy to analyze the perception evidence to determine a context of a user of the computer system, wherein; the control component uses information from the analysis component regarding at least one feature about which to collect data to select the times at which to gather perception evidence from one or more of the plurality of sensors; the analysis component generates the information based on a value of the at least one feature for determining the context; the analysis policy comprises at least one selected from the group consisting of;
a random selection perception policy that randomly selects which features to utilize on a frame by frame basis, a rate-based perception policy that defines observational frequencies and duty cycles for at least one feature or an EVI-based perception policy that determines an expected value of information via a cost-benefit analysis means utilizing at least expected values and cost of analysis values for at least one feature, the cost of analysis values comprising at least one value proportional to an impact to the computing resources employed by the perception system. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 80)
-
-
43. An analysis system comprising at least one processor, the analysis system comprising
a control component that selectively gathers perception evidence to limit utilization of computing resources by a perception system; - and
an analysis component that utilizes an analysis policy to analyze, with the at least one processor, the perception evidence obtained for employment in the perception system;
whereinthe analysis component is interactive with the control component for perception evidence analysis operations, the analysis policy comprises an EVI-based perception policy that determines an expected value of information via a cost-benefit analysis means utilizing at least expected values and cost of analysis values for at least one feature, the cost-benefit analysis means comprise at least one utility model that facilitates in analyzing a benefit of determining a value of the at least one feature, the utility model comprises a conditional utility model that alters functionality dependent upon context, and the expected value of information is determined, at least in part, via utilization of Equation (2);
- and
-
44. A method of analyzing data, comprising:
with the at least one processor; obtaining perception evidence for a perception system from a plurality of sensors; determining a context from the perception evidence; analyzing the perception evidence utilizing an analysis policy to determine a perceived system value indicating a value of utilization of computing resources of a computer system by the perception system; and employing the perceived system value to limit utilization of the computing resources by determining at least one sensor from the plurality of sensors to obtain the perception evidence and a time at which to obtain the perception evidence by the at least one sensor, wherein the time is selected based on an intensity of interaction of a user with the computer system. - View Dependent Claims (45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79)
-
81. A data analysis system, comprising:
-
a plurality of sensors to collect a plurality of features as perception evidence; control means to limit utilization of computing resources of a computer system by a perception system by controlling the plurality of sensors to selectively gather the perception evidence obtained via the perception system; and analyzing means to; analyze the perception evidence utilizing an analysis policy to determine context of use of the computer system by one or more users in which the perception system is operating, and guide the control means to limit the utilization of the computing resources by determining which features from the plurality of features to collect as the perception evidence depending on a value of the features for limiting the utilization of the computing resources by the perception system, based on the determined context;
whereinthe control means selects at least one sensor from the plurality of sensors to collect the features.
-
-
82. A data analysis system, comprising:
-
a first component of a perception system that receives a data query relating to data to determine a context in which the perception system is operating so that to limit utilization of computing resources of a computer system by the perception system, wherein the data comprises perception evidence pertaining to at least one feature collected by a plurality of sensors; and a second component that; analyzes, with the at least one processor, volatility versus persistence of observations of at least one state of the data over time to establish reasonableness in timing of at least one reply to the query, and selects at least one sensor from the plurality of sensors to collect the perception evidence pertaining to the at least one feature to limit the utilization of the computing resources based on a value of the at least one feature in the context, wherein selection of the at least one feature depends on an intensity of interaction of a user with the computer system.
-
Specification