WO2020115904A1 - 学習装置、学習方法、および学習プログラム - Google Patents
学習装置、学習方法、および学習プログラム Download PDFInfo
- Publication number
- WO2020115904A1 WO2020115904A1 PCT/JP2018/045113 JP2018045113W WO2020115904A1 WO 2020115904 A1 WO2020115904 A1 WO 2020115904A1 JP 2018045113 W JP2018045113 W JP 2018045113W WO 2020115904 A1 WO2020115904 A1 WO 2020115904A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- learning
- imitation
- reward
- policy
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Ceased
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N10/00—Quantum computing, i.e. information processing based on quantum-mechanical phenomena
- G06N10/60—Quantum algorithms, e.g. based on quantum optimisation, quantum Fourier or Hadamard transforms
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N7/00—Computing arrangements based on specific mathematical models
- G06N7/01—Probabilistic graphical models, e.g. probabilistic networks
Definitions
- the present invention relates to a learning device, a learning method, and a learning program for learning a model applied to a device that performs processing using a specific model.
- a next-generation computer (hereinafter referred to as a quantum computer) that uses the quantum mechanical principle has been proposed as a device that realizes a large amount of calculation at high speed.
- a quantum computer is a computer that uses quantum effects to perform calculations, and uses the characteristics of superposition to perform calculations at high speed.
- the annealing machine is a computer that operates with a method (quantum annealing method) specialized for combinatorial optimization problems, and operates by giving parameters of a calculation model that operates internally.
- a calculation model is the Ising model, and a device that performs calculation based on such a specific model will be referred to as a specialized device hereinafter.
- Patent Document 1 describes a method for improving the accuracy and calculation time of automatic image recognition by implementing an association graph and a quantum processor.
- the image matching problem is formulated as a quadratic unconstrained binary optimization problem, and the formulated problem is mapped to the quantum processor to operate the quantum processor.
- Specialized devices can process at high speed, but it is necessary to design the objective function in a format that can be solved by the specialized device according to the problem to be solved. Therefore, when using quantum annealing or specialized devices, the analyst needs to express the real-world problem to be solved by a specific physical model (for example, Ising model).
- a specific physical model for example, Ising model
- an object of the present invention is to provide a learning device, a learning method, and a learning program capable of directly learning a specific model used in a quantum computer from the acquired data.
- a learning device is a learning device that learns a model applied to a device that performs processing using a specific model, and includes an input unit that receives a functional input of a reward and imitation learning based on learning data. And the imitation learning unit that learns a policy according to, and the imitation learning unit learns the reward function according to the functional form of the input reward by the imitation learning.
- a learning method is a learning method for learning a model applied to a device that performs processing using a specific model, in which a computer accepts an input of a functional form of reward, and the computer is based on the learning data. It is characterized in that the computer learns the policy by the imitation learning, and the computer learns the reward function according to the functional form of the input reward by the imitation learning.
- a learning program is a learning program applied to a computer that learns a model applied to a device that performs processing using a specific model, and the input processing that receives an input of a functional form of reward in the computer, And imitation learning processing for learning a policy by imitation learning based on learning data, and in the imitation learning processing, the reward function according to the functional form of the input reward is learned by the imitation learning. ..
- a specific model used in a quantum computer can be directly learned from the acquired data.
- FIG. 1 is a block diagram showing an embodiment of an information processing system including a learning device according to the present invention. It is a flowchart which shows the operation example of a learning device. It is a block diagram which shows the outline of the learning device by this invention. It is a schematic block diagram which shows the structure of the computer which concerns on at least 1 embodiment.
- FIG. 1 is a block diagram showing an embodiment of an information processing system 1 including a learning device 100 according to the present invention.
- the information processing system 1 of this embodiment includes a learning device 100 and a device 200.
- the device 200 is a quantum computer that operates based on a specific model.
- the case where the device 200 is a quantum annealing machine will be described as an example.
- the aspect of the device 200 is not limited to the quantum annealing machine.
- the learning device 100 includes a storage unit 10, an input unit 20, an imitation learning unit 30, and an output unit 40.
- the learning device 100 learns a specific model for causing the device 200 to perform processing.
- learning data the state vector
- an environment having a plurality of possible states hereinafter referred to as a target environment
- an agent a person capable of performing a plurality of actions in the environment
- the state vector s may be simply referred to as the state s.
- an agent is an autonomous vehicle.
- the target environment is represented as a set of states of the autonomous vehicle and its surroundings (for example, a map of the surroundings, positions and speeds of other vehicles, and road conditions).
- the action that the agent should take depends on the state of the target environment. In the example of the autonomous vehicle described above, if there is an obstacle ahead, it is necessary to proceed so as to avoid the obstacle. In addition, it is necessary to change the traveling speed of the vehicle according to the condition of the road surface in front and the distance between the vehicle and the vehicle in front.
- a function that outputs the action that the agent should take according to the state of the target environment is called a policy.
- the imitation learning unit 30 described below generates a policy by imitation learning. If the policy is learned to be ideal, the policy will output the optimal action that the agent should take according to the state of the target environment.
- the input unit 20 receives an input of a functional form of reward generated by the imitation learning unit 30 described later.
- the functional form of the reward is a specific model for causing the device 200 to perform processing.
- the functional form of the reward is an Ising model.
- an Ising model will be described as an example of a function form specialized for quantum annealing.
- the functional form of the reward received by the input unit 20 is not limited to the Ising model.
- the input unit 20 inputs the binary neural network or the Hubbard model as the functional form of the reward. May be accepted.
- the imitation learning unit 30 performs imitation learning using data (that is, learning data) in which the state vector s and the action a are associated with each other, and outputs a policy.
- the policy obtained by imitation learning imitates given learning data.
- a policy that is a rule for an agent to select an action is represented by ⁇
- the probability of selecting the action a in the state s under this policy ⁇ is represented by ⁇ (s,a).
- the method of the imitation learning unit 30 performing imitation learning is arbitrary, and the imitation learning unit 30 may output the policy by performing imitation learning using a general method.
- the imitation learning unit 30 performs imitation learning and outputs a reward function. Specifically, the imitation learning unit 30 determines a policy in which the reward r(s) obtained by inputting the state vector s into the reward function r is the input of the function. That is, the action a obtained from the policy is defined by Equation 1 exemplified below.
- the imitation learning unit 30 may formulate the policy as a functional of the reward function. By performing imitation learning using the policy formulated as described above, the imitation learning unit 30 can learn the reward function while learning the policy.
- the probability of selecting the state s′ from the certain state s and the action a can be expressed as ⁇ (a
- the relationship of Expression 2 exemplified below can be defined using the reward function r(s,a). It should be noted that the reward function r (s, a), and is sometimes referred to as a r a (s).
- the imitation learning unit 30 learns the reward function according to the functional form of the input reward by the imitation learning.
- the functional form of the Ising model input to the quantum annealing machine is represented by the following Equation 4.
- (s, a) representing the state corresponds to ⁇ in the Ising model
- the parameter ⁇ corresponds to J ij and h i .
- the imitation learning unit 30 learns the reward function (specifically, the parameter ⁇ ) based on the functional form of the Ising model shown in Expression 4 above. By learning the reward function using such a function form, the imitation learning unit 30 can generate the reward function of the type required in the quantum annealing machine.
- the imitation learning unit 30 learns the reward function by imitation learning based on the functional form of the Ising model.
- the input function form is not limited to the Ising model.
- the imitation learning unit 30 may learn the reward function by imitation learning based on the functional form of the Hubbard model.
- the output unit 40 outputs the reward function generated as a result of the learning of the imitation learning unit 30.
- the generated reward function does not need to be retransformed because it is generated as a model of the type required by the device 200. Therefore, the output unit 40 may directly input (set) the parameter specified by the generated reward function to the device 200.
- the output unit 40 may also output the evaluation result for the generated reward function.
- the method of evaluating the reward function is arbitrary.
- the output unit 40 may output, for example, the closeness between the expert data used for learning (learning data) and the prediction result based on the learned reward function, as the evaluation result for each reward function.
- the input unit 20, the imitation learning unit 30, and the output unit 40 include a computer processor that operates according to a program (learning program) (for example, CPU (Central Processing Unit), GPU (Graphics Processing Unit), FPGA (field-programmable). gate array))).
- a program for example, CPU (Central Processing Unit), GPU (Graphics Processing Unit), FPGA (field-programmable). gate array)
- the program may be stored in the storage unit 10 of the learning device 100, and the processor may read the program and operate as the input unit 20, the imitation learning unit 30, and the output unit 40 according to the program.
- the function of the learning device 100 may be provided in the SaaS (Software as a Service) format.
- the input unit 20, the imitation learning unit 30, and the output unit 40 may be realized by dedicated hardware. Further, some or all of the constituent elements of each device may be realized by a general-purpose or dedicated circuit, a processor, or a combination thereof. These may be configured by a single chip, or may be configured by a plurality of chips connected via a bus. Some or all of the components of each device may be realized by a combination of the above-described circuits and the like and a program.
- the plurality of information processing devices, circuits, etc. may be centrally arranged or distributed. It may be arranged.
- the information processing device, the circuit, and the like may be realized as a form in which a client server system, a cloud computing system, and the like are connected to each other via a communication network.
- the storage unit 10 is realized by, for example, a magnetic disk or the like.
- FIG. 2 is a flowchart showing an operation example of the learning device 100 of this embodiment.
- the input unit 20 receives a functional input of reward (step S11).
- the imitation learning unit 30 learns a policy by imitation learning based on the learning data (step S12). Further, the imitation learning unit 30 learns the reward function according to the functional form of the input reward by the imitation learning (step S13).
- the output unit 40 outputs the reward function generated as a result of learning (step S14).
- the input unit 20 receives the input of the functional form of the reward
- the imitation learning unit 30 learns the policy by imitation learning based on the learning data. Further, the imitation learning unit 30 learns the reward function according to the functional form of the input reward by the imitation learning. Therefore, a specific model used in the quantum computer can be learned from the acquired data.
- the input unit 20 receives the input of the functional form of the reward, and the imitation learning unit 30 learns the reward function according to the functional form of the input reward.
- the Ising model reward function can be obtained by setting and learning the type of the reward function specialized for quantum annealing (for example, Ising model function form). Therefore, since it is not necessary to reconvert the objective function into the Ising model, it is possible to reduce the cost required for the work as a whole.
- the imitation learning unit 30 directly learns the reward function according to the model used by the quantum annealing machine from the learning data to maximize the performance of the quantum annealing machine, various users can use the quantum annealing machine.
- FIG. 3 is a block diagram showing an outline of the learning device according to the present invention.
- the learning device 80 according to the present invention is a learning device (for example, the learning device 100) that learns a model applied to a device (for example, the device 200) that performs processing using a specific model, and has a functional form of reward (
- an input unit 81 for example, the input unit 20
- an imitation learning unit 82 for example, the imitation learning unit 30
- learns a policy by imitation learning based on learning data Is equipped with.
- the imitation learning unit 82 learns the reward function according to the functional form of the input reward by the imitation learning.
- the imitation learning unit 82 may learn the reward function by formulating the policy as a functional of the reward function and performing the imitation learning using the formulated policy.
- the learning device 80 may be a learning device that learns a model applied to a device that performs quantum annealing based on the Ising model (for example, a quantum annealing machine).
- the input unit 81 may accept the functional form of the Ising model
- the imitation learning unit 82 may learn the reward function according to the functional form of the Ising model by imitation learning.
- the imitation learning unit 82 learns a policy by imitation learning based on learning data in which states and behaviors are associated with each other, and sets spins in the Ising model (for example, ⁇ in Expression 4) as states of the learning data.
- the reward function may be learned by determining the parameters of the Ising model (for example, J ij and h i in Expression 4) in association with the action.
- FIG. 4 is a schematic block diagram showing the configuration of a computer according to at least one embodiment.
- the computer 1000 includes a processor 1001, a main storage device 1002, an auxiliary storage device 1003, and an interface 1004.
- the above-described learning device 80 is installed in the computer 1000.
- the operation of each processing unit described above is stored in the auxiliary storage device 1003 in the form of a program (learning program).
- the processor 1001 reads the program from the auxiliary storage device 1003, expands it in the main storage device 1002, and executes the above processing according to the program.
- the auxiliary storage device 1003 is an example of a non-transitory tangible medium.
- Other examples of non-transitory tangible media include magnetic disks, magneto-optical disks, CD-ROMs (Compact Disc Read-only memory), DVD-ROMs (Read-only memory), which are connected via the interface 1004.
- a semiconductor memory etc. are mentioned. Further, when this program is distributed to the computer 1000 via a communication line, the computer 1000 that has received the distribution may expand the program in the main storage device 1002 and execute the above processing.
- the program may be for realizing some of the functions described above. Further, the program may be a so-called difference file (difference program) that realizes the function described above in combination with another program already stored in the auxiliary storage device 1003.
- difference file difference program
- Information Processing System 10 Storage Unit 20 Input Unit 30 Imitation Learning Unit 40 Output Unit 100 Learning Device 200 Device
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Computational Mathematics (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Condensed Matter Physics & Semiconductors (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Algebra (AREA)
- Probability & Statistics with Applications (AREA)
- Computational Linguistics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
Description
10 記憶部
20 入力部
30 模倣学習部
40 出力部
100 学習装置
200 デバイス
Claims (8)
- 特定のモデルを用いて処理を行うデバイスに適用されるモデルを学習する学習装置であって、
報酬の関数形の入力を受け付ける入力部と、
学習データに基づいて模倣学習により方策を学習する模倣学習部とを備え、
前記模倣学習部は、前記模倣学習により、入力された報酬の関数形に応じた報酬関数を学習する
ことを特徴とする学習装置。 - 模倣学習部は、報酬関数の汎関数として方策を定式化し、定式化された方策を用いて模倣学習を行うことにより、報酬関数を学習する
請求項1記載の学習装置。 - イジングモデルに基づいて量子アニーリングを行うデバイスに適用されるモデルを学習する学習装置であって、
入力部は、イジングモデルの関数形を受け付け、
模倣学習部は、模倣学習により、イジングモデルの関数形に応じた報酬関数を学習する
請求項1または請求項2記載の学習装置。 - 模倣学習部は、状態と行動とを対応付けた学習データに基づいて模倣学習により方策を学習し、イジングモデルにおけるスピンを前記学習データの状態および行動に対応付けて当該イジングモデルのパラメータを決定することにより報酬関数を学習する
請求項3記載の学習装置。 - 特定のモデルを用いて処理を行うデバイスに適用されるモデルを学習する学習方法であって、
コンピュータが、報酬の関数形の入力を受け付け、
前記コンピュータが、学習データに基づいて模倣学習により方策を学習し、
前記コンピュータが、前記模倣学習により、入力された報酬の関数形に応じた報酬関数を学習する
ことを特徴とする学習方法。 - コンピュータが、報酬関数の汎関数として方策を定式化し、定式化された方策を用いて模倣学習を行うことにより、報酬関数を学習する
請求項5記載の学習方法。 - 特定のモデルを用いて処理を行うデバイスに適用されるモデルを学習するコンピュータに適用される学習プログラムであって、
前記コンピュータに、
報酬の関数形の入力を受け付ける入力処理、および、
学習データに基づいて模倣学習により方策を学習する模倣学習処理を実行させ、
前記模倣学習処理で、前記模倣学習により、入力された報酬の関数形に応じた報酬関数を学習させる
ための学習プログラム。 - コンピュータに、
模倣学習処理で、報酬関数の汎関数として方策を定式化させ、定式化された方策を用いて模倣学習を行うことにより、報酬関数を学習させる
請求項7記載の学習プログラム。
Priority Applications (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US17/296,798 US12265893B2 (en) | 2018-12-07 | 2018-12-07 | Learning device, learning method, and learning program |
| PCT/JP2018/045113 WO2020115904A1 (ja) | 2018-12-07 | 2018-12-07 | 学習装置、学習方法、および学習プログラム |
| JP2020558788A JP7111178B2 (ja) | 2018-12-07 | 2018-12-07 | 学習装置、学習方法、および学習プログラム |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/JP2018/045113 WO2020115904A1 (ja) | 2018-12-07 | 2018-12-07 | 学習装置、学習方法、および学習プログラム |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2020115904A1 true WO2020115904A1 (ja) | 2020-06-11 |
Family
ID=70973476
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/JP2018/045113 Ceased WO2020115904A1 (ja) | 2018-12-07 | 2018-12-07 | 学習装置、学習方法、および学習プログラム |
Country Status (3)
| Country | Link |
|---|---|
| US (1) | US12265893B2 (ja) |
| JP (1) | JP7111178B2 (ja) |
| WO (1) | WO2020115904A1 (ja) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2022016279A (ja) * | 2020-07-08 | 2022-01-21 | Tdk株式会社 | 成膜システム、工場システム及びウェハの成膜方法 |
| WO2022137506A1 (ja) * | 2020-12-25 | 2022-06-30 | 日本電気株式会社 | 運転評価システム、学習装置、評価結果出力装置、方法およびプログラム |
| WO2023214582A1 (ja) | 2022-05-02 | 2023-11-09 | 三菱重工業株式会社 | 学習装置、学習方法及び学習プログラム |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2017091272A (ja) * | 2015-11-11 | 2017-05-25 | 日本電信電話株式会社 | 内部基準推定装置、方法、及びプログラム |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2143044A4 (en) | 2007-04-19 | 2012-08-08 | Dwave Sys Inc | SYSTEMS, METHODS AND DEVICES FOR AUTOMATIC IMAGE RECOGNITION |
| DE112017002604T5 (de) | 2016-06-21 | 2019-02-21 | Sri International | Systeme und Verfahren für das maschinelle Lernen unter Verwendung eines vertrauenswürdigen Modells |
| US11468334B2 (en) * | 2018-06-19 | 2022-10-11 | International Business Machines Corporation | Closed loop model-based action learning with model-free inverse reinforcement learning |
-
2018
- 2018-12-07 WO PCT/JP2018/045113 patent/WO2020115904A1/ja not_active Ceased
- 2018-12-07 JP JP2020558788A patent/JP7111178B2/ja active Active
- 2018-12-07 US US17/296,798 patent/US12265893B2/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2017091272A (ja) * | 2015-11-11 | 2017-05-25 | 日本電信電話株式会社 | 内部基準推定装置、方法、及びプログラム |
Non-Patent Citations (1)
| Title |
|---|
| UCHIDA, SHIHORI ET AL.: "Estimationof the change of agent's behavior strategy using state-action history.", IEICE TECHNICAL REPORT., vol. 116, no. 521, 6 March 2017 (2017-03-06), pages 7 - 12, ISSN: 0913-5685 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2022016279A (ja) * | 2020-07-08 | 2022-01-21 | Tdk株式会社 | 成膜システム、工場システム及びウェハの成膜方法 |
| JP7633046B2 (ja) | 2020-07-08 | 2025-02-19 | Tdk株式会社 | 成膜システム及びウェハの成膜方法 |
| WO2022137506A1 (ja) * | 2020-12-25 | 2022-06-30 | 日本電気株式会社 | 運転評価システム、学習装置、評価結果出力装置、方法およびプログラム |
| JPWO2022137506A1 (ja) * | 2020-12-25 | 2022-06-30 | ||
| EP4250272A4 (en) * | 2020-12-25 | 2024-01-17 | NEC Corporation | Driving assessment system, learning device, assessment result output device, method, and program |
| JP7552727B2 (ja) | 2020-12-25 | 2024-09-18 | 日本電気株式会社 | 運転評価システム、学習装置、評価結果出力装置、方法およびプログラム |
| WO2023214582A1 (ja) | 2022-05-02 | 2023-11-09 | 三菱重工業株式会社 | 学習装置、学習方法及び学習プログラム |
Also Published As
| Publication number | Publication date |
|---|---|
| US12265893B2 (en) | 2025-04-01 |
| JP7111178B2 (ja) | 2022-08-02 |
| JPWO2020115904A1 (ja) | 2021-09-30 |
| US20210398019A1 (en) | 2021-12-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US12585275B1 (en) | Loss scaling for neural networks | |
| US10902347B2 (en) | Rule creation using MDP and inverse reinforcement learning | |
| US11640518B2 (en) | Method and apparatus for training a neural network using modality signals of different domains | |
| CN109643383B (zh) | 域分离神经网络 | |
| US11200357B2 (en) | Predicting target characteristic data | |
| CN112633310A (zh) | 具有改进的训练鲁棒性地对传感器数据进行分类的方法和系统 | |
| JP7502972B2 (ja) | プルーニング管理装置、プルーニング管理システム及びプルーニング管理方法 | |
| CN112784954A (zh) | 确定神经网络的方法和装置 | |
| WO2020101948A1 (en) | Dynamic precision scaling at epoch granularity in neural networks | |
| CN113591573A (zh) | 多任务学习深度网络模型的训练及目标检测方法、装置 | |
| CN111899594A (zh) | 用于自动驾驶车辆的动态模型的自动化训练数据提取方法 | |
| US20160092789A1 (en) | Category Oversampling for Imbalanced Machine Learning | |
| CN111210016A (zh) | 对包含元素级别操作的神经网络进行剪枝 | |
| JP7226696B2 (ja) | 機械学習方法、機械学習システム及び非一時的コンピュータ可読記憶媒体 | |
| JP7111178B2 (ja) | 学習装置、学習方法、および学習プログラム | |
| JP7111177B2 (ja) | 学習装置、学習方法、および学習プログラム | |
| JP6992891B2 (ja) | 学習装置、情報処理システム、学習方法、および学習プログラム | |
| CN116663650A (zh) | 深度学习模型的训练方法、目标对象检测方法及装置 | |
| CN116710974A (zh) | 在合成数据系统和应用程序中使用域对抗学习的域适应 | |
| US20250061360A1 (en) | Quantum oracle decomposition for simulating quantum computing systems and applications | |
| US12443678B2 (en) | Stepwise uncertainty-aware offline reinforcement learning under constraints | |
| US12275434B2 (en) | Autonomous driving evaluation using data analysis | |
| US12536447B2 (en) | Feature selection in vertical federated learning | |
| CN118691869A (zh) | 对象检测中不平衡数据的实例感知重复因子采样方法 | |
| US20190354833A1 (en) | Method and system for reducing communication frequency in neural network systems |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 18942196 Country of ref document: EP Kind code of ref document: A1 |
|
| ENP | Entry into the national phase |
Ref document number: 2020558788 Country of ref document: JP Kind code of ref document: A |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 18942196 Country of ref document: EP Kind code of ref document: A1 |
|
| WWG | Wipo information: grant in national office |
Ref document number: 17296798 Country of ref document: US |

