# US 12,008,487 B2 · Inference model optimization > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,008,487 B2 - Document ID: US12008487B2 - Status: Granted - Kind: B2 - Application number: 17/122,774 - Filed: 2020-12-15 - Published: 2022-06-16 - Granted: 2024-06-11 - Earliest claimed: 2020-12-15 - Assignees: International Business Machines Corporation - Applicants: International Business Machines Corporation - Inventors: Li Cao; Ze Ming Zhao; Hong Min; Jing Yan Ma - Web page: https://patentgrep.com/patent/US12008487B2 ## CPC - G06F 16/51: Indexing; Data structures therefor; Storage structures - G06N 5/048: Fuzzy inferencing - G06F 16/54: Browsing; Visualisation therefor - H04L 47/562: {Attaching a time tag to queues} - H04L 47/564: {Attaching a deadline to packets, e.g. earliest due date first} - G06N 20/00: Machine learning - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS - H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M) ## Abstract An approach to optimize performance for large scale inference models. Data in the form of images is received from sensors such as cameras. The data is processed to generate data tags associated with the context of the image and portion the images. Model tags are generated based on data characteristics or user input. The tags and their associated data are added to a time-based queue for delivery to the appropriate inference models. Based on the embedded delivery time and frequency, the portioned images are delivered to the appropriate inference models. ## Claims ### Claim 1 A computer-implemented method for optimizing data delivery to inference models, the computer-implemented method comprising: receiving, by one or more processors, data from one or more sensors; processing, by the one or more processors, the data and generating one or more data tags based on characteristics associated with the data and one or more model tags; adding, by the one or more processors, the one or more data tags and the one or more model tags to a time-based tag queue; parsing, by the one or more processors, the time-based tag queue to identify at least one inference model associated with at least one of the one or more data tags; and transmitting, by the one or more processors, a portion of the at least one of the one or more data tags and the data associated with the portion of the at least one of the one or more data tags from an edge side to the identified at least one inference model. ### Claim 2 (depends on 1) The computer-implemented method of claim 1, wherein the data is images and the one or more sensors are cameras. ### Claim 3 (depends on 1) The computer-implemented method of claim 1, wherein the characteristics comprise people, number of people, faces of people, density of people, facial emotions, geographic locations, animals, inanimate objects, and medical abnormalities. ### Claim 4 (depends on 2) The computer-implemented method of claim 2, wherein the data tags are generated based on a context captured in the images. ### Claim 5 (depends on 1) The computer-implemented method of claim 1, wherein the one or more model tags are generated based on user input or based on a context associated with the data. ### Claim 6 (depends on 1) The computer-implemented method of claim 1, wherein the time-based tag queue is sorted on a delivery time and frequency of delivery basis. ### Claim 7 (depends on 6) The computer-implemented method of claim 6, wherein the delivery time and the frequency of delivery is calculated based on a sum of a number of time window offsets in which the one or more data tags appear, for a predetermined number of time windows, divided by the predetermined number of time windows. ### Claim 8 A computer program product for improving quality of a presentation, the computer program product comprising: one or more non-transitory computer readable storage media and program instructions stored on the one or more non-transitory computer readable storage media, the program instructions comprising: program instructions to receive data from one or more sensors; program instructions to process the data and generating one or more data tags based on characteristics associated with the data and one or more model tags; program instructions to add the one or more data tags and the one or more model tags to a time-based tag queue; program instructions to parse the time-based tag queue to identify at least one inference model associated with at least one of the one or more data tags; and program instructions to transmit a portion of the at least one of the one or more data tags and the data associated with the portion of the at least one of the one or more data tags from an edge side to the identified at least one inference model. ### Claim 9 (depends on 8) The computer program product of claim 8, wherein the data is images and the one or more sensors are cameras. ### Claim 10 (depends on 8) The computer program product of claim 8, wherein the characteristics comprise people, number of people, faces of people, density of people, facial emotions, geographic locations, animals, inanimate objects and medical abnormalities. ### Claim 11 (depends on 9) The computer program product of claim 9, wherein the data tags are generated based on a context captured in the images. ### Claim 12 (depends on 8) The computer program product of claim 8, wherein the one or more model tags are generated based on user input or based on a context associated with the data. ### Claim 13 (depends on 8) The computer program product of claim 8, wherein the time-based tag queue is sorted on a delivery time and frequency of delivery basis. ### Claim 14 (depends on 13) The computer program product of claim 13, wherein the delivery time and the frequency of delivery is calculated based on a sum of a number of time window offsets in which the one or more data tags appear, for a predetermined number of time windows, divided by the predetermined number of time windows. ### Claim 15 A computer system for improving quality of a presentation, the computer system comprising: one or more computer processors; one or more computer readable storage media; and program instructions stored on the one or more computer readable storage media for execution by at least one of the one or more processors, the program instructions comprising: program instructions to receive data from one or more sensors; program instructions to process the data and generating one or more data tags based on characteristics associated with the data and one or more model tags; program instructions to add the one or more data tags and the one or more model tags to a time-based tag queue; program instructions to parse the time-based tag queue to identify at least one inference model associated with at least one of the one or more data tags; and program instructions to transmit a portion of the at least one of the one or more data tags and the data associated with the portion of the at least one of the one or more data tags from an edge side to the identified at least one inference model. ### Claim 16 (depends on 15) The computer system of claim 15, wherein the data is images and the one or more sensors are cameras. ### Claim 17 (depends on 16) The computer system of claim 16, wherein the data tags are generated based on a context captured in the images. ### Claim 18 (depends on 15) The computer system of claim 15, wherein the one or more model tags are generated based on user input or based on a context associated with the data. ### Claim 19 (depends on 15) The computer system of claim 15, wherein the time-based tag queue is sorted on a delivery time and frequency of delivery basis. ### Claim 20 (depends on 19) The computer system of claim 19, wherein the delivery time and the frequency of delivery is calculated based on a sum of a number of time window offsets in which the one or more data tags appear, for a predetermined number of time windows, divided by the predetermined number of time windows.