Patent Application
20200250608
The field relates generally to information processing systems, and more particularly to techniques for multi-modal data processing in such systems.
Organizations, companies, and other such enterprises often organize trainings, workshops, seminars, etc. to educate and/or inform employees and other participants about various matters and content. Such types of events are typically organized at non-trivial expense, with respect to both enterprise resources and employee/participant time. With resources and time invested, it is important that enterprises are able to optimize such events by, for example, utilizing high-quality instructors to lead the events. Identifying high-quality instructors, under conventional approaches, typically includes obtaining participant feedback captured in manual review forms. However, participants do not always provide sufficient feedback in such forms, and some details of the event and/or instructor may not be captured in such forms.
Illustrative embodiments of the disclosure provide techniques for evaluations performed by applying machine learning and artificial intelligence techniques to multi-modal data. An exemplary computer-implemented method includes determining one or more audio attributes associated with an instruction event by applying one or more machine learning techniques to audio data captured in connection with the instruction event, determining one or more video attributes associated with the instruction event by applying one or more machine learning techniques to video data captured in connection with the instruction event, and determining one or more image attributes associated with the instruction event by applying one or more machine learning techniques to image data captured in connection with the instruction event. Such a method also includes determining one or more context-based attributes associated with the instruction event by generating and processing audience responses to one or more dynamic context-based audience queries in connection with the instruction event, wherein the one or more dynamic context-based audience queries are generated via applying one or more artificial intelligence techniques to at least a portion of one or more of the audio data, the video data, and the image data. Further, such a method includes generating an evaluation score attributed to at least one instructor of the instruction event based at least in part on the one or more audio attributes, the one or more video attributes, the one or more image attributes, and the one or more context-based attributes, and outputting the evaluation score to at least one of one or more users and one or more centralized platforms.
Illustrative embodiments can provide significant advantages relative to conventional event feedback generation techniques. For example, challenges associated with insufficient and/or incomplete feedback are overcome through automatic monitoring and processing of various forms of multi-modal data inputs throughout an event to facilitate a comprehensive analysis of the instructor and/or the event.
These and other illustrative embodiments described herein include, without limitation, methods, apparatus, systems, and computer program products comprising processor-readable storage media.
Illustrative embodiments will be described herein with reference to exemplary computer networks and associated computers, servers, network devices or other types of processing devices. It is to be appreciated, however, that the invention is not restricted to use with the particular illustrative network and device configurations shown. Accordingly, the term “computer network” as used herein is intended to be broadly construed, so as to encompass, for example, any system comprising multiple networked processing devices.
The user devices 102 may comprise, for example, mobile telephones, laptop computers, tablet computers, desktop computers or other types of computing devices. Such devices are examples of what are more generally referred to herein as “processing devices.” Some of these processing devices are also generally referred to herein as “computers.”
The user devices 102 in some embodiments comprise respective computers associated with a particular company, organization or other enterprise. In addition, at least portions of the computer network 100 may also be referred to herein as collectively comprising an “enterprise network.” Numerous other operating scenarios involving a wide variety of different types and arrangements of processing devices and networks are possible, as will be appreciated by those skilled in the art.
Also, it is to be appreciated that the term “user” in this context and elsewhere herein is intended to be broadly construed so as to encompass, for example, human, hardware, software or firmware entities, as well as various combinations of such entities.
The network 104 is assumed to comprise a portion of a global computer network such as the Internet, although other types of networks can be part of the computer network 100, including a wide area network (WAN), a local area network (LAN), a satellite network, a telephone or cable network, a cellular network, a wireless network such as a Wi-Fi or WiMAX network, or various portions or combinations of these and other types of networks. The computer network 100 in some embodiments therefore comprises combinations of multiple different types of networks, each comprising processing devices configured to communicate using internet protocol (IP) or other related communication protocols.
Additionally, the evaluation system 105 can have an associated database 106 configured to store multi-modal event-related data 107, which may comprise, for example, audio data, video data, image data, and/or context-based quiz data associated with one or more events (such as lectures, presentations, etc.).
The database 106 in the present embodiment is implemented using one or more storage systems associated with the evaluation system 105. Such storage systems can comprise any of a variety of different types of storage including network-attached storage (NAS), storage area networks (SANs), direct-attached storage (DAS) and distributed DAS, as well as combinations of these and other storage types, including software-defined storage.
Also associated with the evaluation system 105 are input-output devices 108, which illustratively comprise keyboards, displays or other types of input-output devices in any combination. Such input-output devices can be used, for example, to support one or more user interfaces to the evaluation system 105, as well as to support communication between the evaluation system 105 and other related systems and devices not explicitly shown.
The evaluation system 105 in the
More particularly, the evaluation system 105 in this embodiment each can comprise a processor 120 coupled to a memory 122 and a network interface 124.
The processor 120 illustratively comprises a microprocessor, a microcontroller, an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other type of processing circuitry, as well as portions or combinations of such circuitry elements. The memory 122 illustratively comprises random access memory (RAM), read-only memory (ROM) or other types of memory, in any combination. The memory 122 and other memories disclosed herein may be viewed as examples of what are more generally referred to as “processor-readable storage media” storing executable computer program code or other types of software programs.
One or more embodiments include articles of manufacture, such as computer-readable storage media. Examples of an article of manufacture include, without limitation, a storage device such as a storage disk, a storage array or an integrated circuit containing memory, as well as a wide variety of other types of computer program products. The term “article of manufacture” as used herein should be understood to exclude transitory, propagating signals.
The network interface 124 allows the evaluation system 105 to communicate over the network 104 with the user devices 102, and illustratively comprises one or more conventional transceivers.
The processor 120 further comprises a machine learning data processor 130, a context-based question-answer generator 132, and an evaluation score algorithm module 134.
It is to be appreciated that this particular arrangement of modules 130, 132 and 134 illustrated in the processor 120 of the
At least portions of the machine learning data processor 130, context-based question-answer generator 132, and evaluation score algorithm module 134 may be implemented at least in part in the form of software that is stored in memory 122 and executed by processor 120.
It is to be understood that the particular set of elements shown in
An exemplary process utilizing machine learning data processor 130, context-based question-answer generator 132, and evaluation score algorithm module 134 of an example evaluation system 105 in computer network 100 will be described in more detail with reference to the flow diagrams of
At least one embodiment includes assessing the quality of an instructor by providing a mathematical index derived from mathematical logic which will indicate a feedback score of the instructor based on various parameters. As detailed herein, such an embodiment includes capturing data-driven feedback with respect to audio data, video data and image data. As also noted, such multi-modal data can be processed within the framework of one or more factors such as, for example, instructor confidence, instructor engagement with audience, instructor correctness, instructor pitch, instructor punctuality, instructor friendliness, content verification, completion, instructor answering of questions, etc.
Additionally, in one or more embodiments, a factor weight system is implemented wherein predefined weights (for example, 3, 2, and 1) are applied to the modes (i.e., audio, video, and image) of captured and/or processed data (whose values are calculated through one or more algorithms, as further described herein). Such weightages are determined and updated in connection with one or more machine learning models.
Accordingly, at least one embodiment includes generating instructor quality values by calculating the feedback score of the instructor. By way merely of example, a final value may be in the form of a percentage (out of 100%) value. Further, one or more embodiments can include additional features such as implementing a dynamic dashboard that leverages the generated quality values to provide insights about instructors and help instructors improve in one or more targeted areas. Such additional features can also include implementing one or more machine learning techniques of logistic regression to select an instructor for a future event based on generated quality values and various variables/parameters desired and/or required in an instructor for the given event.
As detailed herein, one or more embodiments include utilizing and processing audio data. Specifically, such embodiments include processing factors such as the tone of the instructor, the pitch of the instructor, the correctness of the instructor in answering questions and/or teaching, understanding context of audience participation, the sentiment of the instructor, etc. The processing of audio data can be carried out, for example, by audio feedback algorithms as well as machine learning compression algorithms such as Dual Ask-Answer Network (DAANET) modelling techniques for machine reading comprehension (MRC) related analysis. At least one embodiment includes obtaining an audio transcript from the audio file in question and analyzing the content of the transcript to identify various types of entities talking during the event. Such an embodiment can further include determining an understanding the context and content of what is being taught/discussed during the event, and identifying when a question is asked and answered by training one or more artificial intelligence models applied to the relevant audio files.
One or more embodiments also include generating questions and answers using the audio files captured using, for example, a two-stage synthesis network. The correctness of what is being taught or discussed can then be checked, as well as whether the questions were answered properly by the instructor using a DAANet model for MRC.
As also detailed herein, one or more embodiments include utilizing and processing video data. Specifically, such embodiments include processing factors such as the sentiment of the instructor, the sentiment of the audience, and the confidence of the audience. The processing of video data can be carried out, for example, by video feedback algorithms including one or more dynamic and/or recognition algorithms, which can be utilized to calculate values for one or more factors of the video data (such as noted above).
As further detailed herein, one or more embodiments include utilizing and processing image data. Specifically, such embodiments include processing factors such as the confidence of the instructor and/or audience at various stages of the event. The processing of video data can be carried out, for example, by image feedback algorithms including, for example, one or more convolutional neural network- (CNN-) based image classification algorithms. At least one embodiment includes capturing random-timed photos, during the event, of the audience and the instructor to create a time frame of images. Using such a created time frame of images, such an embodiment further includes calculating confidence values attributed to the instructor and the audience (using one or more recognition algorithms) at various stages of time during the event (thus creating a journey and/or experience of the instructor and the audience). Such actions can additionally be expanded for various factors other than confidence.
Much of the above-detailed multi-modal factor data analysis can include training models based on previous labeled data. Outputs can include, by way of example, percentage-based values among the audience group (e.g., 20% of the audience looks angry, etc.). Also, in one or more embodiments, additional factors can be added.
Additionally, as also detailed herein (and further described in connection with
Also, one or more embodiments include analyzing the user's answers to help determine whether the instructor was able to portray the concept(s) properly to the audience. Further, in at least one embodiment, noise is a factor which will help in identifying how organized and/or interactive the event was.
Additionally, as noted above, one or more embodiments include implementing weights for the multi-modal data and/or factors thereof. In such an embodiment, each mode of data (i.e., audio, video, image) and/or each factor has a predefined weight based, for example, on user-defined levels of importance of the data and/or factor.
In the above algorithms, N defines the noise element factor in the event/session, Wi=[1, 2, 3], which represents the predefined weights for audio, video and images data factors, Vi=[1, 2, 3], which represents the determined value(s) of the audio, video and image data factors, Wj=[4, . . . , n], which represents the predefined weights for the context-based quiz data factor, Vj=[4, . . . , n], which represents the determined value(s) of the context-based quiz data factor, and “n” signifies future factors.
The IFC algorithm 412 provides an image data score to evaluation score algorithm module 134, the VFC algorithm 414 provides a video data score to evaluation score algorithm module 134, and the AFC algorithm 416 provides an audio data score to evaluation score algorithm module 134. Additionally, the context-based question-answer generator 132, based on analysis of participant responses against dynamically generated questions, provides a context-based quiz data score to evaluation score algorithm module 134. Based on such inputs, the evaluation score algorithm module 134 generates a final feedback instructor score 418, which is then uploaded onto a centralized platform 420 for instructor profiles. Such a platform 420, for example, can be used by an organization to select particular instructors for particular events based on the profiles contained therein.
Accordingly, one or more embodiments include autonomously leveraging generated profiles of various user devices (such as mobile devices, watches, laptops, etc.) based upon the external environment (a public venue, an outdoor venue, a user's home, a user's work location, in a meeting or event, etc.). At least one embodiment includes creating meaningful context information from a given environment by performing topic modelling and segregating the meaningful information from noise. Such an embodiment can thereby create artificial intelligence-based quizzes in real-time without the need for a static pre-existing question-answer database related to the given topic. Additionally, and as further described herein, such an embodiment is extensible to various domains based at least in part on the ability to gather and process inputs via various digital formats.
Referring to
Referring to
Additionally, in at least one embodiment, the context-based question-answer generator 132 is capable of receiving inputs via multiple platforms. For example, if the setting is a classroom setup, input can be received by an array of connected microphones and/or video cameras. By way of further example, if the setting is a resume checker, input can be received in the form of a digital curriculum vitae (CV). Once such input is received, the input can be sent (as shown in step 604 of
At the end of the above-noted stage, an assorted list of Q/A pairs is generated, from which the context-based question-answer generator 132 can form a quiz (as depicted in
The context-based question-answer generator 132, which receives various context-related information as input, dynamically generates and outputs a set of well-defined Q/A pairs based on the context-related information. In one or more embodiments, the dynamic generation of such Q/A pairs includes implementing one or more machine reading comprehension-based artificial intelligence models. As depicted in
In this embodiment, the process includes steps 800 through 810. These steps are assumed to be performed by the processor 120 utilizing its modules 130, 132 and 134.
Step 800 includes determining one or more audio attributes associated with an instruction event by applying one or more machine learning techniques to audio data captured in connection with the instruction event. The one or more audio attributes can include tone of the at least one instructor, pitch of the at least one instructor, sentiment of the at least one instructor, and/or correctness of the at least one instructor in answering one or more audience queries. Additionally, determining the one or more audio attributes can include training at least one model associated with the one or more machine learning techniques using labeled historical audio data.
Step 802 includes determining one or more video attributes associated with the instruction event by applying one or more machine learning techniques to video data captured in connection with the instruction event. The one or more video attributes can include sentiment of the at least one instructor and/or sentiment of the audience. Also, determining the one or more video attributes can include training at least one model associated with the one or more machine learning techniques using labeled historical video data.
Step 804 includes determining one or more image attributes associated with the instruction event by applying one or more machine learning techniques to image data captured in connection with the instruction event. The one or more image attributes can include sentiment of the at least one instructor and/or sentiment of the audience. Additionally, determining the one or more image attributes can include training at least one model associated with the one or more machine learning techniques using labeled historical image data.
Step 806 includes determining one or more context-based attributes associated with the instruction event by generating and processing audience responses to one or more dynamic context-based audience queries in connection with the instruction event, wherein the one or more dynamic context-based audience queries are generated via applying one or more artificial intelligence techniques to at least a portion of one or more of the audio data, the video data, and the image data. The one or more context-based attributes can include attentiveness of the audience and/or comprehension of instruction event content by the audience.
Step 808 includes generating an evaluation score attributed to at least one instructor of the instruction event based at least in part on the one or more audio attributes, the one or more video attributes, the one or more image attributes, and the one or more context-based attributes. Generating the evaluation score can include applying a predefined weight to each of the one or more audio attributes, the one or more video attributes, the one or more image attributes, and the one or more context-based attributes. Step 810 includes outputting the evaluation score to at least one of one or more users and one or more centralized platforms.
In this embodiment, the process includes steps 900 through 908. These steps are assumed to be performed by utilizing module 132.
Step 900 includes obtaining multi-modal data from one or more data sources, wherein the multi-modal data pertains to a given event. The multi-modal data includes audio data, video data, image data, and/or text data. Step 902 includes removing one or more items of noise from the obtained multi-modal data by applying one or more filtering techniques to the multi-modal data, thereby generating filtered multi-modal data. Step 904 includes creating a comprehensive set of multi-modal data pertaining to at least a portion of the given event by aggregating the filtered multi-modal data in accordance with one or more topic modelling techniques and removing any items of duplicate filtered multi-modal data.
Step 906 includes dynamically generating one or more question-answer pairs related to the given event by applying one or more machine reading comprehension-based artificial intelligence models to the comprehensive set of multi-modal data. Dynamically generating the one or more question-answer pairs can include dynamically generating question-answer pairs of varying difficulty. Additionally, dynamically generating the one or more question-answer pairs can include utilizing context information pertaining to the one or more participants of the given event, wherein the context information includes a type of device being used by the one or more participants, etc.
Step 908 includes outputting at least a portion of the one or more questions from the one or more question-answer pairs to one or more participants of the given event. At least one embodiment also includes calculating a confidence value attributed to each answer provided by the one or more participants in response to the one or more questions. Calculating the confidence value attributed to each answer provided by the one or more participants can include applying one or more similarity matching algorithms to each answer provided by the one or more participants and the dynamically generated answers corresponding to the one or more output questions. Additionally, such an embodiment can also include determining one or more features attributed to the one or more participants of the given event based at least in part on the calculated confidence values, wherein the features include attentiveness of the one or more participants, comprehension by the one or more participants, etc.
Accordingly, the particular processing operations and other functionality described in conjunction with the flow diagrams of
The above-described illustrative embodiments provide significant advantages relative to conventional approaches. For example, some embodiments are configured to automatically process various forms of multi-modal data inputs throughout an event to facilitate a comprehensive analysis of the instructor and/or the event. These and other embodiments can effectively enable selection of an instructor based at least in part on outputs from machine learning techniques of logistic regression with respect to various features which need to be present for a particular event.
It is to be appreciated that the particular advantages described above and elsewhere herein are associated with particular illustrative embodiments and need not be present in other embodiments. Also, the particular types of information processing system features and functionality as illustrated in the drawings and described above are exemplary only, and numerous other arrangements may be used in other embodiments.
As mentioned previously, at least portions of the information processing system 100 may be implemented using one or more processing platforms. A given such processing platform comprises at least one processing device comprising a processor coupled to a memory. The processor and memory in some embodiments comprise respective processor and memory elements of a virtual machine or container provided using one or more underlying physical machines. The term “processing device” as used herein is intended to be broadly construed so as to encompass a wide variety of different arrangements of physical processors, memories and other device components as well as virtual instances of such components. For example, a “processing device” in some embodiments can comprise or be executed across one or more virtual processors. Processing devices can therefore be physical or virtual and can be executed across one or more physical or virtual processors. It should also be noted that a given virtual device can be mapped to a portion of a physical one.
Some illustrative embodiments of a processing platform that may be used to implement at least a portion of an information processing system comprises cloud infrastructure including virtual machines implemented using a hypervisor that runs on physical infrastructure. The cloud infrastructure further comprises sets of applications running on respective ones of the virtual machines under the control of the hypervisor. It is also possible to use multiple hypervisors each providing a set of virtual machines using at least one underlying physical machine. Different sets of virtual machines provided by one or more hypervisors may be utilized in configuring multiple instances of various components of the system.
These and other types of cloud infrastructure can be used to provide what is also referred to herein as a multi-tenant environment. One or more system components, or portions thereof, are illustratively implemented for use by tenants of such a multi-tenant environment.
As mentioned previously, cloud infrastructure as disclosed herein can include cloud-based systems such as Amazon Web Services (AWS), Google Cloud Platform (GCP) and Microsoft Azure. Virtual machines provided in such systems can be used to implement at least portions of one or more of a computer system in illustrative embodiments. These and other cloud-based systems in illustrative embodiments can include object stores such as Amazon S3, GCP Cloud Storage, and Microsoft Azure Blob Storage.
In some embodiments, the cloud infrastructure additionally or alternatively comprises a plurality of containers implemented using container host devices. For example, as detailed herein, a given container of cloud infrastructure illustratively comprises a Docker container or other type of Linux Container (LXC). The containers may run on virtual machines in a multi-tenant environment, although other arrangements are possible. The containers may be utilized to implement a variety of different types of functionality within the system 100. For example, containers can be used to implement respective processing devices providing compute and/or storage services of a cloud-based system. Again, containers may be used in combination with other virtualization infrastructure such as virtual machines implemented using a hypervisor.
Illustrative embodiments of processing platforms will now be described in greater detail with reference to
The cloud infrastructure 1000 further comprises sets of applications 1010-1, 1010-2, . . . 1010-L running on respective ones of the VMs/container sets 1002-1, 1002-2, . . . 1002-L under the control of the virtualization infrastructure 1004. The VMs/container sets 1002 may comprise respective VMs, respective sets of one or more containers, or respective sets of one or more containers running in VMs. In some implementations of the
An example of a hypervisor platform that may be used to implement a hypervisor within the virtualization infrastructure 1004 is the VMware® vSphere® which may have an associated virtual infrastructure management system such as the VMware® vCenter™. The underlying physical machines may comprise one or more distributed processing platforms that include one or more storage systems.
In other implementations of the
As is apparent from the above, one or more of the processing modules or other components of system 100 may each run on a computer, server, storage device or other processing platform element. A given such element may be viewed as an example of what is more generally referred to herein as a “processing device.” The cloud infrastructure 1000 shown in
The processing platform 1100 in this embodiment comprises a portion of system 100 and includes a plurality of processing devices, denoted 1102-1, 1102-2, 1102-3, . . . 1102-K, which communicate with one another over a network 1104.
The network 1104 may comprise any type of network, including by way of example a global computer network such as the Internet, a WAN, a LAN, a satellite network, a telephone or cable network, a cellular network, a wireless network such as a Wi-Fi or WiMAX network, or various portions or combinations of these and other types of networks.
The processing device 1102-1 in the processing platform 1100 comprises a processor 1110 coupled to a memory 1112.
The processor 1110 may comprise a microprocessor, a microcontroller, an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other type of processing circuitry, as well as portions or combinations of such circuitry elements.
The memory 1112 may comprise random access memory (RAM), read-only memory (ROM) or other types of memory, in any combination. The memory 1112 and other memories disclosed herein should be viewed as illustrative examples of what are more generally referred to as “processor-readable storage media” storing executable program code of one or more software programs.
Articles of manufacture comprising such processor-readable storage media are considered illustrative embodiments. A given such article of manufacture may comprise, for example, a storage array, a storage disk or an integrated circuit containing RAM, ROM or other electronic memory, or any of a wide variety of other types of computer program products. The term “article of manufacture” as used herein should be understood to exclude transitory, propagating signals. Numerous other types of computer program products comprising processor-readable storage media can be used.
Also included in the processing device 1102-1 is network interface circuitry 1114, which is used to interface the processing device with the network 1104 and other system components, and may comprise conventional transceivers.
The other processing devices 1102 of the processing platform 1100 are assumed to be configured in a manner similar to that shown for processing device 1102-1 in the figure.
Again, the particular processing platform 1100 shown in the figure is presented by way of example only, and system 100 may include additional or alternative processing platforms, as well as numerous distinct processing platforms in any combination, with each such platform comprising one or more computers, servers, storage devices or other processing devices.
For example, other processing platforms used to implement illustrative embodiments can comprise different types of virtualization infrastructure, in place of or in addition to virtualization infrastructure comprising virtual machines. Such virtualization infrastructure illustratively includes container-based virtualization infrastructure configured to provide Docker containers or other types of LXCs.
As another example, portions of a given processing platform in some embodiments can comprise converged infrastructure such as VxRail™, VxRack™, VxBlock™, or Vblock® converged infrastructure commercially available from VCE, the Virtual Computing Environment Company, now the Converged Platform and Solutions Division of Dell EMC.
It should therefore be understood that in other embodiments different arrangements of additional or alternative elements may be used. At least a subset of these elements may be collectively implemented on a common processing platform, or each such element may be implemented on a separate processing platform.
Also, numerous other arrangements of computers, servers, storage products or devices, or other components are possible in the information processing system 100. Such components can communicate with other elements of the information processing system 100 over any type of network or other communication media.
For example, particular types of storage products that can be used in implementing a given storage system of a distributed processing system in an illustrative embodiment include VNX® and Symmetrix VMAX® storage arrays, software-defined storage products such as ScaleIO™ and ViPR®, all-flash and hybrid flash storage arrays such as Unity™, cloud storage products such as Elastic Cloud Storage (ECS), object-based storage products such as Atmos®, scale-out all-flash storage arrays such as XtremIO™, and scale-out NAS clusters comprising Isilon® platform nodes and associated accelerators, all from Dell EMC. Combinations of multiple ones of these and other storage products can also be used in implementing a given storage system in an illustrative embodiment.
It should again be emphasized that the above-described embodiments are presented for purposes of illustration only. Many variations and other alternative embodiments may be used. Also, the particular configurations of system and device elements and associated processing operations illustratively shown in the drawings can be varied in other embodiments. Thus, for example, the particular types of processing platforms, modules, cloud-based systems and resources deployed in a given embodiment and their respective configurations may be varied. Moreover, the various assumptions made above in the course of describing the illustrative embodiments should also be viewed as exemplary rather than as requirements or limitations of the disclosure. Numerous other alternative embodiments within the scope of the appended claims will be readily apparent to those skilled in the art.
