Mattdl/ContinualPrototypeEvolution
Codebase for Continual Prototype Evolution (CoPE) to attain perpetually representative prototypes for online and non-stationary datastreams. Includes implementation of the Pseudo-Prototypical Proxy (PPP) loss.
This project helps machine learning engineers and researchers build robust AI models that can continuously learn from new, evolving data without forgetting previously learned information. It takes in streaming, non-stationary labeled data and produces a constantly updated model capable of classifying new inputs accurately over time. It's designed for those developing AI systems that must adapt to real-world data drift.
No commits in the last 6 months.
Use this if you need to train a classification model on data that changes over time, and your model must learn continuously without explicit task boundaries or 'retraining from scratch' on new information.
Not ideal if your data is static and well-defined upfront, or if you can easily retrain your models periodically on all available historical data.
Stars
44
Forks
7
Language
Python
License
—
Category
Last pushed
Mar 21, 2022
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/Mattdl/ContinualPrototypeEvolution"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
aimagelab/mammoth
An Extendible (General) Continual Learning Framework based on Pytorch - official codebase of...
LAMDA-CL/PyCIL
PyCIL: A Python Toolbox for Class-Incremental Learning
GMvandeVen/continual-learning
PyTorch implementation of various methods for continual learning (XdG, EWC, SI, LwF, FROMP, DGR,...
LAMDA-CL/LAMDA-PILOT
🎉 PILOT: A Pre-trained Model-Based Continual Learning Toolbox
mmasana/FACIL
Framework for Analysis of Class-Incremental Learning with 12 state-of-the-art methods and 3 baselines.