Target Molecule Specification Using Explored and Unexplored Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for training determination models for molecule application properties are resource-intensive, time-consuming, and often require cumbersome trial-and-error processes, making them inefficient and less suitable for new applications.
Innovation Solution
A method that utilizes a trained determination model parameterized with both explored and unexplored data sets to determine molecule properties, allowing for more effective and objective training, reducing resource consumption and increasing model applicability across various contexts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional trial-and-error training processes are used to develop determination models, then the model can be trained to achieve high accuracy in a specific application context, but the process becomes time-consuming and consumes high computational and laboratory resources
Solution Approach 1:
The patent applies preliminary action by pre-defining the training data set based on the unexplored data set before the actual training process. The unexplored data set is used to determine which explored molecules should be included in the training data set, thereby preparing the training process in advance and avoiding time-consuming trial-and-error iterations during training.
Solution Approach 2:
The patent uses copying by creating a virtual representation of the training process through the unexplored data set. Instead of physically performing multiple training iterations with different data sets, the system creates a computational model that simulates the training process, allowing for efficient selection of training molecules without repeated physical experimentation.
2Measurement precision
If traditional trial-and-error training processes are used to develop determination models, then the model can be trained to achieve high accuracy in a specific application context, but the process consumes high computational and laboratory resources
Solution Approach 1:
The patent applies preliminary action by pre-defining the training data set based on the unexplored data set before the actual training process. The unexplored data set is used to determine which explored molecules should be included in the training data set, thereby preparing the training process in advance and avoiding time-consuming trial-and-error iterations during training.
Solution Approach 2:
The patent uses copying by creating a virtual representation of the training process through the unexplored data set. Instead of physically performing multiple training iterations with different data sets, the system creates a computational model that simulates the training process, allowing for efficient selection of training molecules without repeated physical experimentation.
3Adaptability or versatility
If a completely new determination model is trained for each new application, then the model can be optimized for that specific application, but the time and computational resource intensive training process must be repeated
Solution Approach 1:
The patent applies universality by creating a framework where the same determination model can be adapted to multiple applications through the use of unexplored data sets. The unexplored data set serves as a universal template that can be applied to different application contexts, allowing the model to be reconfigured for new applications without requiring complete retraining from scratch.
Solution Approach 2:
The patent applies preliminary action by pre-defining the training data set based on the unexplored data set before the actual training process. The unexplored data set is used to determine which explored molecules should be included in the training data set, thereby preparing the training process in advance and avoiding time-consuming trial-and-error iterations during training.
4Reliability
If traditional training processes are used, then the model can be trained based on historical molecule data, but the models become less suitable for new applications requiring completely new training processes
Solution Approach 1:
The patent applies dynamics by making the training data set flexible and adaptable rather than fixed. The unexplored data set serves as a dynamic template that can be adjusted according to the specific application context, allowing the training data set to evolve and adapt to new applications while maintaining the core structure of the determination model.
Solution Approach 2:
The patent applies universality by creating a framework where the same determination model can be adapted to multiple applications through the use of unexplored data sets. The unexplored data set serves as a universal template that can be applied to different application contexts, allowing the model to be reconfigured for new applications without requiring complete retraining from scratch.
Data Source
AI summary
The invention refers to a method for determining a target molecule. A target property and a digital representation of a potential target molecule is provided. Then a model is utilized for determining a property of a potential target molecule. The model has been parameterized based on an explored data set and an unexplored data set. The explored data set comprises a property for a plurality of explored molecules and molecule characterizing parameter values for the plurality of explored molecules. The unexplored data set comprising characterizing parameter values for a plurality of unexplored molecules. The determined property of the potential target molecule is compared with the target property. Based on the comparison, either i) the potential target molecule is determined as the target molecule, or ii) a new potential molecule is determined and the determination of the property repeated. The determined target molecule is then provided.