Treatment Selection Engine Using Cost Probability Distributions

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Machine learning models in online systems often fail to balance exploration and exploitation, leading to suboptimal treatment selection for users, and underestimate treatment costs due to assuming static costs without considering uncertainty, resulting in reduced value for the online system.

Innovation Solution

An online concierge system employs a user treatment engine that generates treatment cost predictions and scores using probability distributions, adjusting for user data and treatment costs to select treatments that balance rewards and costs, and updates parameters based on application results.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If machine learning models rely on treatments with known rewards, then the reliability of treatment selection is improved, but the productivity of the online system deteriorates due to suboptimal treatment selection

Engineering Contradiction:

Improvetreatment selection reliabilityVSAvoidonline system value

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

The patent changes the parameter representation from static cost estimates to probability distributions (e.g., normal distributions with mean and standard deviation). This allows the model to capture uncertainty in treatment costs and rewards, enabling better balancing of exploration and exploitation by representing both the expected value and the variability of treatment outcomes.

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent introduces dynamic updating of treatment selection parameters based on observed outcomes. The probability distributions are updated as new data becomes available, allowing the system to adaptively balance exploration of less-known treatments with exploitation of well-performing treatments, thereby improving both reliability and productivity.

Inventive Principle:

Principle #15Dynamics

2Device complexity

If machine learning models assume static treatment costs, then the device complexity is reduced, but the measurement precision of treatment costs deteriorates due to ignoring cost uncertainty

Engineering Contradiction:

Improvemodel complexityVSAvoidtreatment cost estimation accuracy

Core Design Contradiction:

Device complexityVSMeasurement precision

Solution Approach 1:

The patent transforms the static cost parameter into a probability distribution with multiple parameters (mean and standard deviation). This change allows the model to represent both the expected cost and the uncertainty around that cost, significantly improving measurement precision while adding manageable complexity through the use of standard statistical distributions.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS12198173B2Treatment selection and application using machine learning to predict cost of treatment where information about cost of treatment is unknown

Publication Date: 2025.01.14 MAPLEBEAR INC

AI summary

A user treatment engine uses user data describing characteristics of a user to evaluate a set of treatments that the user treatment engine may apply to the user. The user treatment engine generates treatment cost predictions for the treatments and generates treatment scores for the set of treatments based on the treatment cost predictions for the treatments and the user data for the user. The user treatment engine selects and applies a treatment from the set of treatments based on the generated treatment scores. The user treatment engine determines a reward to the online concierge system for the application of the treatment to the user and updates treatment selection parameters for the applied treatment based on the determined reward.