# Jackson Burns

Published articles for Jackson Burns.

This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles.

## How to predict the behavior of molecules from scant data

DevFeed: [How to predict the behavior of molecules from scant data](<https://devfeed.tech/articles/how-to-predict-the-behavior-of-molecules-from-scant-data-121787.md>)

Original publisher: [Read original article](<https://news.mit.edu/2026/how-to-predict-molecule-behavior-with-scant-data-1106>)

Author: David L. Chandler | Department of Chemical Engineering

Published: 2026-11-06T19:00:00Z

Content type: article

Language: en

Sources: [MIT AI News](<https://devfeed.tech/sources/mit-ai-news.md>)

Topics: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Training AI Models](<https://devfeed.tech/topics/training-ai-models.md>), [few-shot](<https://devfeed.tech/topics/few-shot.md>), [Neural Network](<https://devfeed.tech/topics/neural-network.md>), [Open Source Models & Datasets](<https://devfeed.tech/topics/open-source-models-datasets.md>), [generalization in machine learning](<https://devfeed.tech/topics/generalization-in-machine-learning.md>)

Tags: [ai-in-drug-discovery](<https://devfeed.tech/tags/ai-in-drug-discovery.md>), [akshat-shirish-zalte](<https://devfeed.tech/tags/akshat-shirish-zalte.md>), [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [charlles-abreu](<https://devfeed.tech/tags/charlles-abreu.md>), [chemeleon](<https://devfeed.tech/tags/chemeleon.md>), [chemical-engineering](<https://devfeed.tech/tags/chemical-engineering.md>), [chemistry](<https://devfeed.tech/tags/chemistry.md>), [chemprop](<https://devfeed.tech/tags/chemprop.md>), [deep-learning-models](<https://devfeed.tech/tags/deep-learning-models.md>), [department-of-energy-doe](<https://devfeed.tech/tags/department-of-energy-doe.md>), [doe-funding](<https://devfeed.tech/tags/doe-funding.md>), [drug-discovery](<https://devfeed.tech/tags/drug-discovery.md>), [drugs](<https://devfeed.tech/tags/drugs.md>), [foundation-model](<https://devfeed.tech/tags/foundation-model.md>), [foundation-models](<https://devfeed.tech/tags/foundation-models.md>), [jackson-burns](<https://devfeed.tech/tags/jackson-burns.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [mit-cheme](<https://devfeed.tech/tags/mit-cheme.md>), [molecular-descriptors](<https://devfeed.tech/tags/molecular-descriptors.md>), [molecule-properties](<https://devfeed.tech/tags/molecule-properties.md>), [national-institutes-of-health-nih](<https://devfeed.tech/tags/national-institutes-of-health-nih.md>), [pubchem](<https://devfeed.tech/tags/pubchem.md>), [research](<https://devfeed.tech/tags/research.md>), [school-of-engineering](<https://devfeed.tech/tags/school-of-engineering.md>), [william-green](<https://devfeed.tech/tags/william-green.md>)

### AI overview

CheMeleon uses a pre-trained neural network to help predict molecular properties when only a few task-specific data points are available. It learns from descriptors--such as molecular weight and oxygen count--for one million molecules in PubChem, then can be fine-tuned for properties including solubility, viscosity, bioactivity, or reactivity. Unlike classical expert-designed models, it learns general chemical patterns rather than relying on specialists to specify which descriptors matter for a property. The team reports that CheMeleon outperformed other foundation models and classical models across 58 test cases, potentially reducing the number of molecules that must be synthesized and tested. It is open source through ChemProp. The excerpt does not provide detailed performance metrics for individual test cases.

### Source excerpt

A machine-learning approach called CheMeleon provides a practical way to develop new drugs or other molecules with much less detailed, expensive experimentation.