11 results about "Network approach" patented technology

Approaches presented herein provide for unsupervised domain transfer learning. In particular, three neural networks can be trained together using at least labeled data from a first domain and unlabeled data from a second domain. Features of the data are extracted using a feature extraction network. A first classifier network uses these features to classify the data, while a second classifier network uses these features to determine the relevant domain. A combined loss function is used to optimize the networks, with a goal of the feature extraction network extracting features that the first classifier network is able to use to accurately classify the data, but prevent the second classifier from determining the domain for the image. Such optimization enables object classification to be performed with high accuracy for either domain, even though there may have been little to no labeled training data for the second domain.

Approaches presented herein provide systems and methods for content generation systems that incorporate a denoising diffusion generative adversarial network (DDGAN) into a content generation pipeline. A generator associated with the DDGAN may be conditioned on a set of input images that include at least a noisy image from a diffusion engine, an upsampled low resolution image, and a historical image. The generator may be used to generate an output image having one or more properties that are different from a content engine. Weights for the generator may be determined during a training process that includes a discriminator that evaluates at least the noisy image, the upsampled low resolution image, and a noised image produced from the output image.

Approaches are described herein for secure masking of channel activity for transmission security (TRANSEC) enabled terminals in satellite communication networks. Embodiments include techniques for masking channel activity at the link-layer in the outroute and / or in the inroute. For example, in the outroute, embodiments encapsulate packet data units (PDUs) using a TRANSEC-compatible stream encapsulation protocol. In the inroute, embodiments encapsulate burst transmissions using a TRANSEC-compatible inroute burst encapsulation protocol. Outroute and / or inroute activity can be further masked by creating the impression of constant-rate traffic, constant-rate bandwidth allocations, and the like.

Approaches are described herein for link-layer address concealment for transmission security (TRANSEC) enabled (TE) terminals in satellite communication networks. For example, TE terminals can randomly select a proposed masked system assigned identifier (mSAI), which is a temporary identifier different from any real identifiers of the terminal (e.g., a real SAI, real electronic serial number, etc.). The ground station (e.g., inroute group manager) checks availability of the mSAI. If available, the request is acknowledged, the mSAI is assigned to the terminal, and bandwidth is allocated to the terminal using the mSAI. Subsequent allocated bursts can be sent by the terminal using the allocated bandwidth based on the mSAI. Some embodiments include techniques for handling requests for an unavailable mSAI, simultaneous requests by different terminals for the same mSAI, periodic changes of mSAIs for added security, and use of mSAI-related timers to support additional features.

Approaches presented herein provide for the attestation of devices in managed networks, in order to verify state and establish trust in those devices as well as in the managed network and / or subnets. The devices in a network, including devices such as network switches, can perform self-attestation by transmitting attestation evidence using one or more network messages. One or more verifiers can verify the attestation evidence and determine which devices or subnets are trusted. Data and messages can then be routed along trusted paths through a trusted network or subnet, such that all devices along those paths are trusted devices. In order to reduce the volume of attestation traffic for large networks, a network device can provide attestation evidence to a verifier that is connected to that device, rather than propagating all evidence for all devices to a single verifier. Once verified, a list of trusted devices can be propagated rather than the instances of evidence that were used for the verification.

The invention discloses a wind instrument matching method based on reinforcement learning under multiple agents. The method comprises the following steps: extracting music features according to an original MIDI project file, and constructing a music feature model; constructing an expert data set according to the music feature model, and constructing a pre-training model according to the expert data set; constructing a wind instrument dictionary according to a wind instrument method, and constructing a multi-agent reinforcement learning model training environment; constructing a multi-agent reinforcement learning model; training a multi-agent reinforcement learning network in a mode of constructing a target network; according to the invention, the music features can be more comprehensively sorted, and the sorted music features are applied to the automatic matching device of the wind instrument.

Approaches are described herein for link-layer address concealment for transmission security (TRANSEC) enabled (TE) terminals in satellite communication networks. For example, TE terminals can randomly select a proposed masked system assigned identifier (mSAI), which is a temporary identifier different from any real identifiers of the terminal (e.g., a real SAI, real electronic serial number, etc.). The ground station (e.g., inroute group manager) checks availability of the mSAI. If available, the request is acknowledged, the mSAI is assigned to the terminal, and bandwidth is allocated to the terminal using the mSAI. Subsequent allocated bursts can be sent by the terminal using the allocated bandwidth based on the mSAI. Some embodiments include techniques for handling requests for an unavailable mSAI, simultaneous requests by different terminals for the same mSAI, periodic changes of mSAIs for added security, and use of mSAI-related timers to support additional features.

The invention provides a networking method and system based on a vehicle-mounted terminal, belongs to the field of automobiles, and aims to receive related signals of a whole vehicle in real time and seamlessly, noninductively and losslessly access the vehicle-mounted terminal to the Internet through a smart phone. Existing application programs (such as navigation, online music, a voice assistant and an application store) on the vehicle-mounted system can be networked in the mode without any modification, non-inductive networking which can be used as soon as getting on the vehicle is finally achieved on user experience, a user does not need to enter a setting menu of a smart phone to manually open a WiFi hotspot or other network sharing modes, and the user experience is improved. Compatibility to all network protocols is realized functionally, and the possibility of a high-end networking function is provided for a low-cost vehicle machine commercially.

Approaches are described herein for securely initializing a TRANSEC-enabled terminal (TET) in a satellite communication network. At a secure location, a unique Terminal Master Key and Electronic Serial Number, are securely burned into the terminal's memory. Upon booting, the terminal verifies its identity using a signed TE file and encrypted keys. When deployed, the terminal initially sends an unallocated burst with a fake ESN to obtain bandwidth allocation (e.g., from an inroute bandwidth allocator). The terminal proceeds with an initial association request using a randomly generated identifier, followed by a transport-layer-secure registration process (e.g., with a network management system over HTTPS). After successful registration, the TET receives link-layer key materials and shifts to using its real ESN with link-layer security (LLS) for subsequent communications.

This invention discloses a wind farm fault voltage recovery method based on topology reconstruction. The proposed method utilizes a radial network for both normal operation and fault isolation of the wind farm. Topology-level optimization calculations enable the wind farm to achieve both the low redundancy and high reliability of traditional radial and ring networks. Compared to the traditional radial network approach where circuit breakers directly disconnect during internal faults, the proposed topology reconstruction method ensures both the recovery of voltage at the wind turbine generator terminals and the continued power supply from downstream wind turbines without disconnecting from the grid.

The invention provides a task allocation method based on an improved DQN algorithm, and belongs to the field of task allocation, and the method comprises the steps: creating a DQN algorithm model, creating a training model, generating scene parameters of a current scene, defining an agent classification standard, carrying out the pairing of agents and actions through combining a task allocation rule, and calculating a local reward and a global reward. The method comprises the steps of obtaining a reward value used for DQN network updating, updating a node value of an evaluation network based on the reward value used for DQN network updating to obtain an updated evaluation network, updating parameters of a target network based on the updated evaluation network to enable the target network to approach the evaluation network, and obtaining a model after current scene optimization. According to the method, different rules and decision strategies are automatically selected in a dynamic environment based on an improved deep Q learning algorithm of the rules, so that the system can continuously adjust decisions according to real-time scene information, and thus a more efficient task allocation decision can be made in the face of changeable actual scenes.