17768082. OPTIMIZATION APPARATUS, OPTIMIZATION METHOD AND NON-TRANSITORY COMPUTER-READABLE MEDIUM STORING OPTIMIZATION PROGRAM simplified abstract (NEC Corporation)
Contents
- 1 OPTIMIZATION APPARATUS, OPTIMIZATION METHOD AND NON-TRANSITORY COMPUTER-READABLE MEDIUM STORING OPTIMIZATION PROGRAM
- 1.1 Organization Name
- 1.2 Inventor(s)
- 1.3 OPTIMIZATION APPARATUS, OPTIMIZATION METHOD AND NON-TRANSITORY COMPUTER-READABLE MEDIUM STORING OPTIMIZATION PROGRAM - A simplified explanation of the abstract
- 1.4 Simplified Explanation
- 1.5 Potential Applications
- 1.6 Problems Solved
- 1.7 Benefits
- 1.8 Potential Commercial Applications
- 1.9 Possible Prior Art
- 1.10 Original Abstract Submitted
OPTIMIZATION APPARATUS, OPTIMIZATION METHOD AND NON-TRANSITORY COMPUTER-READABLE MEDIUM STORING OPTIMIZATION PROGRAM
Organization Name
Inventor(s)
OPTIMIZATION APPARATUS, OPTIMIZATION METHOD AND NON-TRANSITORY COMPUTER-READABLE MEDIUM STORING OPTIMIZATION PROGRAM - A simplified explanation of the abstract
This abstract first appeared for US patent application 17768082 titled 'OPTIMIZATION APPARATUS, OPTIMIZATION METHOD AND NON-TRANSITORY COMPUTER-READABLE MEDIUM STORING OPTIMIZATION PROGRAM
Simplified Explanation
The present invention relates to an optimization apparatus for achieving highly accurate optimization for a submodular function without the need for preparing data for machine learning. The apparatus includes a determination unit, a reward acquisition unit, a calculation unit, and an update unit to optimize policies in an objective function with diminishing marginal utility and convexity.
- Determination unit: Identifies one or more execution policies from a predetermined policy in the objective function.
- Reward acquisition unit: Obtains the reward, which is the execution result in the objective function for the determined execution policy.
- Calculation unit: Computes the update rate of the policy based on the reward.
- Update unit: Adjusts the policy based on the calculated update rate.
Potential Applications
The technology can be applied in various fields such as finance, logistics, and resource allocation where accurate optimization of submodular functions is required.
Problems Solved
1. Achieving highly accurate optimization for submodular functions without the need for data preparation. 2. Optimizing policies in an objective function with diminishing marginal utility and convexity.
Benefits
1. Improved efficiency in optimizing policies. 2. Enhanced accuracy in submodular function optimization. 3. Reduced reliance on data for machine learning.
Potential Commercial Applications
Optimization software for businesses, algorithm development for financial institutions, and resource allocation tools for logistics companies can benefit from this technology.
Possible Prior Art
Prior art in the field of optimization algorithms and submodular function optimization may include research papers, patents, and existing software solutions that address similar problems.
Unanswered Questions
How does the apparatus handle complex optimization problems in real-time scenarios?
The apparatus utilizes a combination of determination, reward acquisition, calculation, and update units to handle complex optimization problems efficiently in real-time scenarios.
What is the scalability of the optimization apparatus for large-scale applications?
The scalability of the optimization apparatus depends on the computational resources available and the complexity of the optimization problem. Further research and testing are needed to determine its scalability for large-scale applications.
Original Abstract Submitted
A purpose of the present invention is to achieve highly accurate optimization for a submodular function when a policy is optimized without preparing data for machine learning. An optimization apparatus () according to the present invention includes a determination unit () that determines one or more execution policies from a predetermined policy in an objective function having diminishing marginal utility and convexity, a reward acquisition unit () that acquires reward, which is an execution result in the objective function for the determined execution policy, a calculation unit () that calculates an update rate of the policy based on the reward, and an update unit () that updates the policy based on the update rate.
