Abstract: Embodiments disclosed herein model lifelong intent detection as a class-incremental learning where a new set of intents/classes are added at each incremental step. To address the issue of catastrophic forgetting during lifelong intent detection (LID), an incremental learner is provided with Prompt Augmented Generative Replay, wherein unlike existing approaches that store real samples in replay memory, only concept words obtained from old intents are stored, which reduces memory consumption and speeds up incremental training still enabling not forgetting the old intents. Joint training of an incremental learner is carried out for LID and a pseudo-labeled utterance generation with objective is to classify a user utterance into one of multiple pre-defined intents by minimizing a total Loss function comprising a LID loss function, a Labeled Utterance Generation loss function, a Supervised Contrastive Training loss function, and a Knowledge Distillation loss function.

Abstract: The embodiments herein provide a method for predicting content based recommendations. The method includes recognizing, by a first electronic device (100A) in a first IoT network (100), content displayed on a display (130A) of the first electronic device (100A). Further, the method includes predicting, by the first electronic device (100A), at least one recommendation corresponding to the content based on a history of user activities performed by a user using at least one of the first electronic device (100A) and other electronic devices (100B-100F) in the first IoT network (100). Further, the method includes providing, by the first electronic device (100A), the at least one recommendation on an action prediction panel of a keyboard of the first electronic device (100A).

Abstract: Existing semi-supervised and unsupervised approaches for intent discovery require an estimate of the number of new intents present in the user logs. The present disclosure receives labeled utterances from known intents and update parameters of a pre-trained language model (PLM). Representation learning and clustering is performed iteratively using labeled and unlabeled utterances from known intents and unlabeled utterances from unknown intents to fine-tune PLM and a plurality of clusters is generated. Cluster merger algorithm is executed iteratively on generated plurality of clusters. A query cluster is obtained by randomly selecting one cluster from the plurality of clusters and by obtaining a corresponding plurality of nearest neighbors based on a cosine-similarity. A response for merging the query cluster and corresponding plurality of nearest neighbors is obtained, and a new cluster is created.

Abstract: Embodiments disclosed herein model lifelong intent detection as a class-incremental learning where a new set of intents/classes are added at each incremental step. To address the issue of catastrophic forgetting during lifelong intent detection (LID), an incremental learner is provided with Prompt Augmented Generative Replay, wherein unlike existing approaches that store real samples in replay memory, only concept words obtained from old intents are stored, which reduces memory consumption and speeds up incremental training still enabling not forgetting the old intents. Joint training of an incremental learner is carried out for LID and a pseudo-labeled utterance generation with objective is to classify a user utterance into one of multiple pre-defined intents by minimizing a total Loss function comprising a LID loss function, a Labeled Utterance Generation loss function, a Supervised Contrastive Training loss function, and a Knowledge Distillation loss function.

Abstract: Existing semi-supervised and unsupervised approaches for intent discovery require an estimate of the number of new intents present in the user logs. The present disclosure receives labeled utterances from known intents and update parameters of a pre-trained language model (PLM). Representation learning and clustering is performed iteratively using labeled and unlabeled utterances from known intents and unlabeled utterances from unknown intents to fine-tune PLM and a plurality of clusters is generated. Cluster merger algorithm is executed iteratively on generated plurality of clusters. A query cluster is obtained by randomly selecting one cluster from the plurality of clusters and by obtaining a corresponding plurality of nearest neighbors based on a cosine-similarity. A response for merging the query cluster and corresponding plurality of nearest neighbors is obtained, and a new cluster is created.

Abstract: A system and method for automated validation testing of software applications during software development and after development stages of the product life lifecycle. The disclosed system uses pre-trained machine learning models and image recognition algorithms to generate test cases and test data from text and images in product specification documents of the software applications.

Abstract: The present disclosure relates to an apparatus for decoding an encoded Unified Audio and Speech stream. The apparatus comprises a core decoder for decoding the encoded Unified Audio and Speech stream. The core decoder includes a fast Fourier transform, FFT, module implementation based on a Cooley-Tuckey algorithm. The FFT module is configured to determine a discrete Fourier transform, DFT. Determining the DFT involves recursively breaking down the DFT into small FFTs based on the Cooley-Tucker algorithm and using radix-4 if a number of points of the FFT is a power of 4 and using mixed radix if the number is not a power of 4. Performing the small FFTs involves applying twiddle factors. Applying the twiddle factors involves referring to pre-computed values for the twiddle factors.

Abstract: The present disclosure relates to an apparatus for decoding an encoded Unified Audio and Speech stream. The apparatus comprises a core decoder for decoding the encoded Unified Audio and Speech stream. The core decoder includes an upmixing unit adapted to perform mono to stereo upmixing. The upmixing unit includes a decorrelator unit D adapted to apply a decorrelation filter to an input signal. The decorrelator unit is adapted to determine filter coefficients for the decorrelation filter by referring to pre-computed values. The present disclosure further relates to a an apparatus for encoding a Unified Audio and Speech stream, as well as to corresponding methods and storage media.

Abstract: The present disclosure relates to an apparatus for decoding an encoded Unified Audio and Speech stream. The apparatus comprises a core decoder for decoding the encoded Unified Audio and Speech stream. The core decoder includes an eSBR unit for extending a bandwidth of an input signal, the eSBR unit including a QMF based harmonic transposer. The QMF based harmonic transposer is configured to process the input signal in the QMF domain, in each of a plurality of synthesis subbands, to extend the bandwidth of the input signal. The QMF based harmonic transposer is configured to operate at least in part based on pre-computed information. The present disclosure further relates to corresponding methods and storage media.

Abstract: A system and method for automated validation testing of software applications during software development and after development stages of the product life lifecycle. The disclosed system uses pre-trained machine learning models and image recognition algorithms to generate test cases and test data from text and images in product specification documents of the software applications.

Abstract: The embodiments herein provide a method for predicting content based recommendations. The method includes recognizing, by a first electronic device (100A) in a first IoT network (100), content displayed on a display (130A) of the first electronic device (100A). Further, the method includes predicting, by the first electronic device (100A), at least one recommendation corresponding to the content based on a history of user activities performed by a user using at least one of the first electronic device (100A) and other electronic devices (100B-100F) in the first IoT network (100). Further, the method includes providing, by the first electronic device (100A), the at least one recommendation on an action prediction panel of a keyboard of the first electronic device (100A).

Abstract: The present disclosure relates to an apparatus for decoding an encoded Unified Audio and Speech stream. The apparatus comprises a core decoder for decoding the encoded Unified Audio and Speech stream. The core decoder includes an eSBR unit for extending a bandwidth of an input signal, the eSBR unit including a QMF based harmonic transposer. The QMF based harmonic transposer is configured to process the input signal in the QMF domain, in each of a plurality of synthesis subbands, to extend the bandwidth of the input signal. The QMF based harmonic transposer is configured to operate at least in part based on pre-computed information. The present disclosure further relates to corresponding methods and storage media.

Abstract: The present disclosure relates to an apparatus for decoding an encoded Unified Audio and Speech stream. The apparatus comprises a core decoder for decoding the encoded Unified Audio and Speech stream. The core decoder includes a fast Fourier transform, FFT, module implementation based on a Cooley-Tuckey algorithm. The FFT module is configured to determine a discrete Fourier transform, DFT. Determining the DFT involves recursively breaking down the DFT into small FFTs based on the Cooley-Tucker algorithm and using radix-4 if a number of points of the FFT is a power of 4 and using mixed radix if the number is not a power of 4. Performing the small FFTs involves applying twiddle factors. Applying the twiddle factors involves referring to pre-computed values for the twiddle factors.

Abstract: The present disclosure relates to an apparatus for decoding an encoded Unified Audio and Speech stream. The apparatus comprises a core decoder for decoding the encoded Unified Audio and Speech stream. The core decoder includes an upmixing unit adapted to perform mono to stereo upmixing. The upmixing unit includes a decorrelator unit D adapted to apply a decorrelation filter to an input signal. The decorrelator unit is adapted to determine filter coefficients for the decorrelation filter by referring to pre-computed values. The present disclosure further relates to a an apparatus for encoding a Unified Audio and Speech stream, as well as to corresponding methods and storage media.

Abstract: To help manage resources in a cellular wireless communication system, a base station may track air interface resource allocation for each UE the base station is serving with a call and may identify one or more of the UEs or calls as candidates for handover from the base station on grounds that air interface resource allocation to each of the identified one or more UEs is high. Upon receipt of a request from a UE to initiate a new call, the base station may then predict based at least on that UE's channel quality what the air interface resource allocation rate will be for the requested call. And if the predicted rate for the given UE is threshold low, the base station may select one of the candidate UEs and force a handover of the selected UE's call from the base station.

Abstract: In accordance with the disclosed methods and systems, while providing a service that enables user equipment devices (UEs) being served by a first network to engage in signaling with a second network via the first network, a controller or other network entity may detect an error condition associated with the service, such as a threshold-high level of signaling or an interface failure. In response to detecting the error condition, base stations in the first network may then direct one or more UEs being served by the first network to refrain from using the given service. As a result, the one or more UEs may refrain from using the given service and may instead use another approach to engage in signaling with the second network, which may enable the one or more UEs to circumvent the error condition until it gets remedied.

Abstract: Systems, methods, and computer-readable media for managing bearers are provided. In a current telecommunications system, a maximum number of bearers are available to simultaneously support various services. By utilizing different internet protocol addresses from differing networks, the maximum number of bearers may be increased. Specifically, by registering a user device with both an IPv4 and IPv6 network, with different IP addresses associated with the respective bearers, a user device may access an additional number of bearers in excess of the previous maximum number of bearers available.

Abstract: A first home packet gateway (PGW) device may receive a binding update message on behalf of a wireless communication device (WCD). The first home PGW device may be associated with a first pool of Internet Protocol (IP) addresses, and the WCD may be assigned a particular IP address. It may be determined that the first pool of IP addresses does not include the particular IP address, and that the particular IP address is included in a second pool of IP addresses that is associated with a second home PGW device. Possibly in response to determining that the particular IP address is included in the second pool of IP addresses that is associated with the second home PGW device, the first home PGW device may forward the binding update message to the second home PGW device.

Abstract: A method is disclosed to help keep a UE's network identifier available for reassignment to that UE even after that UE engages in a particular type of handover. In accordance with the method, a network may make a determination that a UE should engage in a particular type of handover. In response to making this determination, the network may terminate the UE's active connection. However, based on the particular type of handover in which the network determined that the UE should engage, the network may also reserve the network identifier assigned for the UE's active connection.

Abstract: A method and system for responding to a call-teardown request generated, for instance, as a result of a user equipment device (UE) losing coverage of a first serving radio access network (RAN) while the UE is engaged in a call with another party. A signaling intermediary receives the call-teardown message en route from a first call server to a second call server and responsively determines whether the UE is now served by a second RAN. Responsive to determining that the UE is served by the second RAN, the intermediary arranges for establishment of a new call connection with the UE via the second RAN and for linking of the new call connection with an existing leg of the call with the other party, and the intermediary discards the call-teardown message rather than forwarding it along to the second call server.