中科微点-全球专利检索

  1. Login
  2. Sign Up

Search Results

420 records (took 0.044 seconds)

    Systems and methods for collaborative filtering with variational autoencoders
    1.
    发明授权
    Systems and methods for collaborative filtering with variational autoencoders 有权

    公开(公告)号：US12198051B2

    公开(公告)日：2025-01-14

    申请号：US18096198

    申请日：2023-01-12

    Applicant: D-WAVE SYSTEMS INC.

    Inventor： William G. Macready ,  Jason T. Rolfe

    IPC: G06N3/047 ,  G06F18/214 ,  G06N3/045 ,  G06N3/08 ,  G06N10/00

    Abstract: Collaborative filtering systems based on variational autoencoders (VAEs) are provided. VAEs may be trained on row-wise data without necessarily training a paired VAE on column-wise data (or vice-versa), and may optionally be trained via minibatches. The row-wise VAE models the output of the corresponding column-based VAE as a set of parameters and uses these parameters in decoding. In some implementations, a paired VAE is provided which receives column-wise data and models row-wise parameters; each of the paired VAEs may bind their learned column- or row-wise parameters to the output of the corresponding VAE. The paired VAEs may optionally be trained via minibatches. Unobserved data may be explicitly modelled. Methods for performing inference with such VAE-based collaborative filtering systems are also disclosed, as are example applications to search and anomaly detection.

    Systems and methods for hybrid quantum-classical computing
    5.
    发明授权
    Systems and methods for hybrid quantum-classical computing 有权

    公开(公告)号：US11900264B2

    公开(公告)日：2024-02-13

    申请号：US16785125

    申请日：2020-02-07

    Applicant: D-WAVE SYSTEMS INC.

    Inventor： Catherine McGeoch ,  William W. Bernoudy

    IPC: G06N5/01 ,  G06F15/163 ,  G06F17/18 ,  G06N10/00

    CPC classification number: G06N5/01 ,  G06F15/163 ,  G06F17/18 ,  G06N10/00

    Abstract: Hybrid quantum-classical approaches for solving computational problems in which results from a quantum processor are combined with an exact method executed on a classical processor are described. Quantum processors can generate candidate solutions to a combinatorial optimization problem, but since quantum processors can be probabilistic, they are unable to certify that a solution is an optimal solution. A hybrid quantum-classical exact solver addresses this problem by combining outputs from a quantum annealing processor with a classical exact algorithm that is modified to exploit properties of the quantum computation. The exact method executed on a classical processor can be a Branch and Bound algorithm. A Branch and Bound algorithm can be modified to exploit properties of quantum computation including a) the sampling of multiple low-energy solutions by a quantum processor, and b) the embedding of solutions in a regular structure such as a native hardware graph of a quantum processor.

    TOPOLOGICALLY PROTECTED QUBITS, PROCESSORS WITH TOPOLOGICALLY PROTECTED QUBITS, AND METHODS FOR USE OF TOPOLOGICALLY PROTECTED QUBITS
    7.
    发明公开
    TOPOLOGICALLY PROTECTED QUBITS, PROCESSORS WITH TOPOLOGICALLY PROTECTED QUBITS, AND METHODS FOR USE OF TOPOLOGICALLY PROTECTED QUBITS 审中-公开

    公开(公告)号：US20230370069A1

    公开(公告)日：2023-11-16

    申请号：US17883874

    申请日：2022-08-09

    Applicant: D-WAVE SYSTEMS INC.

    Inventor： Mohammad H. Amin ,  Richard G. Harris

    IPC: H03K19/195 ,  G06N10/40

    CPC classification number: H03K19/195 ,  G06N10/40

    Abstract: A logical qubit, a quantum processor, and a method of performing an operation on the logical qubit are discussed. The logical qubit includes first and second tunable couplers and a plurality of fixed couplers, with at least one fixed coupler providing four physical qubit interaction. The first and second tunable couplers and the fixed couplers enforce even parity in any connected qubits. The logical qubit has a plurality of physical qubits with qubits connected to the first tunable coupler and a first fixed coupler, qubits connected to the second tunable coupler and a second fixed coupler, and qubits connected between the first fixed coupler and the second fixed coupler. Each fixed coupler is connected to at least two physical qubits and at least two paths connect the first tunable coupler and the second tunable coupler, with one path communicating with a microwave line.

    SYSTEMS AND METHODS FOR HEURISTIC ALGORITHMS WITH VARIABLE EFFORT PARAMETERS
    8.
    发明公开
    SYSTEMS AND METHODS FOR HEURISTIC ALGORITHMS WITH VARIABLE EFFORT PARAMETERS 审中-公开

    公开(公告)号：US20230316094A1

    公开(公告)日：2023-10-05

    申请号：US18126566

    申请日：2023-03-27

    Applicant: D-WAVE SYSTEMS INC.

    Inventor： Pau Farré Pérez ,  Jack R. Raymond

    IPC: G06N5/01 ,  G06N10/40 ,  G06F9/445

    CPC classification number: G06N5/01 ,  G06N10/40 ,  G06F9/44505

    Abstract: A heuristic solver is wrapped in a meta algorithm that will perform multiple sub-runs within the desired time limit, and expand or reduce the effort based on the time it has taken so far and the time left. The goal is to use the largest effort possible as this typically increases the probability of success. In another implementation, the meta algorithm iterates the time-like parameter from a small value, and determine the next test-value so as to minimize time to target collecting data at large effort only as necessary. The meta algorithm evaluates the energy of the solutions obtained to determine whether to increase or decrease the value of the time-like parameter. The heuristic algorithm may be Simulated Annealing, the heuristic algorithm may run on a quantum processor, including a quantum annealing processor or a gate-model quantum processor.

    SYSTEMS AND METHODS FOR COLLABORATIVE FILTERING WITH VARIATIONAL AUTOENCODERS
    10.
    发明公开
    SYSTEMS AND METHODS FOR COLLABORATIVE FILTERING WITH VARIATIONAL AUTOENCODERS 审中-公开

    公开(公告)号：US20230222337A1

    公开(公告)日：2023-07-13

    申请号：US18096198

    申请日：2023-01-12

    Applicant: D-WAVE SYSTEMS INC.

    Inventor： William G. Macready ,  Jason T. Rolfe

    IPC: G06N3/08 ,  G06N10/00

    CPC classification number: G06N3/08 ,  G06N10/00 ,  G06N3/045 ,  G06F18/2148

    Abstract: Collaborative filtering systems based on variational autoencoders (VAEs) are provided. VAEs may be trained on row-wise data without necessarily training a paired VAE on column-wise data (or vice-versa), and may optionally be trained via minibatches. The row-wise VAE models the output of the corresponding column-based VAE as a set of parameters and uses these parameters in decoding. In some implementations, a paired VAE is provided which receives column-wise data and models row-wise parameters; each of the paired VAEs may bind their learned column- or row-wise parameters to the output of the corresponding VAE. The paired VAEs may optionally be trained via minibatches. Unobserved data may be explicitly modelled. Methods for performing inference with such VAE-based collaborative filtering systems are also disclosed, as are example applications to search and anomaly detection.

Patent Agency Ranking