Juniper Networks, Inc.

Servicenow, Inc.

Quaizar Vohra

Jean-Marc Frailong

Ravi Shekhar

Gunes Aybay

Pradeep S Sindhu

Anjan Venkatramani

Umesh Kondur

Mitul Tiwari

Jeelani Syed

Srivatsava Daruru

Vijayabhaskar Annamalai Kalusivalingam

Aleksandar Luka Ratkovic

Sreekanth Reddy

Madhusudan Mathihalli

Jaihari V Loganathan

Michael Francis O'Gorman

Shesha Bhushan Sreenivasamurthy

Ravi Narasimhan Raj

Kaushik Rangadurai

Deepak Bobbarjung

Abhisaar Yadav

Joseph LaSalle White

Nischal Sheth

Jainendra Kumar

Arijit Sarcar

Ronak Shah

Murali Vemula

Manoj Tiwari

Prasennaram Dhanushkodi

Nishanth Gaddam

Predrag Spasic

Amit Shukla

Anil Kumar Madamala

Arthi R Ayyangar

Xiangwen Xu

Jignesh Parmar

Olivier Nguyen

Neha A Kaushil

Soham Parikh

Ziaul Hasan Hashmi

Orlando Marquez

Patrice Bechard

Shounak Purkayastha

Prabhu Seshachellam

Mishanth Gaddam

A method is provided for efficiently providing sentiments or other manual labels for textual training data. The method includes using an embedding model to project acquired user text to an embedding vector in an embedding space. Distances (e.g., cosine similarities) between this embedding vector and the embedding vectors determined for a plurality of already-label user text training examples are then determined. The already-labeled user text that has the shortest distance is determined and the label thereof is prospectively applied to the acquired user text and presented to a user for approval. The user can approve the prospectively applied label, in which case the newly acquired text is added to the training data with the prospectively applied label associated therewith for later use in training a language model. Alternatively, the user can decline the prospectively applied label and apply an alternative label to the newly acquired text.

Method for filtering and semi-automatically labeling training data

In one embodiment, edge devices can be configured to be coupled to a multi-stage switch fabric and peripheral processing devices. The edge devices and the multi-stage switch fabric can collectively define a single logical entity. A first edge device from the edge devices can be configured to be coupled to a first peripheral processing device from the peripheral processing devices. The second edge device from the edge devices can be configured to be coupled to a second peripheral processing device from the peripheral processing devices. The first edge device can be configured such that virtual resources including a first virtual resource can be defined at the first peripheral processing device. A network management module coupled to the edge devices and configured to provision the virtual resources such that the first virtual resource can be migrated from the first peripheral processing device to the second peripheral processing device.

Methods and apparatus related to a flexible data center security architecture

Example data processing systems and methods are described. In one implementation, a system accesses a corpus of data and analyzes the data contained in the corpus of data to identify multiple documents. The system generates vector indexes for the multiple documents such that the vector indexes allow a computing system to quickly access the plurality of documents and identify an answer to a question associated with the corpus of data.

Generation and utilization of vector indexes for data processing systems and methods

Data processing systems and methods

In one embodiment, an apparatus includes a switch core that has a multi-stage switch fabric. A first set of peripheral processing devices coupled to the multi-stage switch fabric by a set of connections that have a protocol. Each peripheral processing device from the first set of peripheral processing devices is a storage node that has virtualized resources. The virtualized resources of the first set of peripheral processing devices collectively define a virtual storage resource interconnected by the switch core. A second set of peripheral processing devices coupled to the multi-stage switch fabric by a set of connections that have the protocol. Each peripheral processing device from the first set of peripheral processing devices is a compute node that has virtualized resources. The virtualized resources of the second set of peripheral processing devices collectively define a virtual compute resource interconnected by the switch core.

Methods and apparatus related to virtualization of data center resources

In some embodiments, a non-transitory processor-readable medium stores code representing instructions configured to cause a processor to receive, from an access switch, a first signal including forwarding state information associated with a first peripheral processing device from a set of peripheral processing devices. The code can further represent instructions configured to cause the processor to receive, from the first peripheral processing device, a second signal including a data packet. The code can further represent instructions configured to cause the processor to send, to a replication engine associated with the set of peripheral processing devices, a third signal such that the replication engine (1) defines a copy of the data packet, which is included within the third signal, and (2) sends, to a second peripheral processing device from the set of peripheral processing devices, a fourth signal including the copy of the data packet.

Multicasting within distributed control plane of a switch

In some embodiments, a network management module is operatively coupled to a set of edge devices that are coupled to a set of peripheral processing devices. The network management module can receive a signal associated with a broadcast protocol from an edge device from the set of edge devices in response to that edge device being operatively coupled to a switch fabric. The network management module can provision that edge device in response to receiving the signal. The network management module can define multiple network control entities at the set of edge devices such that each network control entity from the multiple network control entities can provide forwarding-state information associated with at least one peripheral processing device from the set of peripheral processing devices to at least one remaining network control entity from the multiple network control entities using a selective protocol.

Methods and apparatus for automatically provisioning resources within a distributed control plane of a switch

In some embodiments, an apparatus includes a compute device to communicate with a network control entity at each access switch from a set of access switches that define a portion of a data plane having a switch fabric coupling as hierarchical peers each access switch from the set of access switches. The compute device is operable to define a portion of a control plane that includes the network control entities from the set of access switches such that the compute device is hierarchically removed from the network control entities from the set of access switches. The compute device is operable to receive forwarding-state information from a first access switch from the set of access switches. The compute device to send the forwarding-state information to a second access switch from the set of access switches.

Methods and apparatus related to a switch fabric system having a multi-hop distributed control plane and a single-hop data plane

In some embodiments, an apparatus includes a scheduler disposed at a control device of a switch fabric system. The scheduler is configured to receive a control plane request associated with the switch fabric system having a data plane and a control plane separate from the data plane. The scheduler is configured to designate a control plane entity based on the control plane request and state information of each control plane entity from a set of control plane entities associated with the control plane and instantiated as a virtual machine. The scheduler is configured to send a signal to a compute device of the switch fabric system in response to the control plane request such that the control plane entity is instantiated as a virtual machine at the compute device.

Methods and apparatus for virtualizing switch control plane engine

Growing community of inventors

Santa Clara, CA, United States of America

Quaizar Vohra