Improve AI: Machine Learning Platform for Ranking and Scoring
Improve AI helps software companies implement recommendation and ranking systems. It is designed for teams requiring on-device decision making for mobile applications.
At a glance
- Category
- Data & Analytics
- Best for
- Software companies, Enterprise companies, Mobile app developers, Machine learning engineers
- Pricing
- The libraries are free under the MIT license. The Tracker/Trainer is free for internal business and personal use under the Sustainable Use License.
- Key use cases
- Recommendation Systems, Personalization, Query Re-ranking, Multi-variate Optimization
- Integrations
- AWS S3, AWS Lambda
- Official website
- improve.ai
Improve AI is a contextual multi-armed bandit platform designed for ranking and scoring items. It combines XGBoost with Thompson Sampling to help models balance the exploration of new options with the use of established options.
The tool is intended for software companies and enterprises implementing recommendation systems, query re-ranking, or personalized user experiences. It provides SDKs for Python, Swift, and Java, which support on-device decision making on iOS and Android to help reduce network latency.
Buyers should note that this is a technical tool. It uses reinforcement learning, where developers assign rewards based on business metrics like revenue or conversions to guide the model's optimization over time.
Key Features
Uses JSON-encodable data as context to help personalize how items are scored and ranked for different users or conditions.
Supports a reward-based training approach where positive or negative outcomes are used to update the model.
Supports scoring and ranking directly on iOS and Android devices.
Designed to balance exploration of novel possibilities and exploitation of known high-reward options.
Provides libraries for Python, Swift, and Java.
Use Cases
Ranking a list of items to present relevant options to a user.
Using contextual data to adjust the scoring of items based on user attributes or environment.
Updating score columns for products in a database to refine search results.
Optimizing multiple variables to support business metrics.
Best For
Integrations
Pricing
The libraries are free under the MIT license. The Tracker/Trainer is free for internal business and personal use under the Sustainable Use License.
FAQ
It is used to create apps that score and rank items, supporting workflows like recommendation systems and personalized user experiences.
The libraries are free under the MIT license, and the Tracker/Trainer is free for internal business and personal use.
It provides SDKs for Python, Swift, and Java, with support for on-device decisions on iOS and Android.
It uses reinforcement learning where developers track interactions and assign rewards based on business outcomes.
Source category: Data & Analytics
Source subcategory: Machine Learning Platform
