AETHER

KO | EN

gitlite — search

#python #java #python3 #arduino #golang #machine-learning #rust #html #flask #javascript #seismology #nodejs

★ 16 Open GitHub ↗

AETHER: AlphaEarth–POI Enriched Representation Learning for Human-Centered Urban Analysis

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# AETHER: AlphaEarth–POI Enriched Representation Learning

AETHER is a lightweight multimodal alignment framework that aligns **AlphaEarth (AE) embeddings** with **POI-derived urban semantics** through contrastive learning.
The aligned representations support both **urban downstream tasks** and **natural language–conditioned spatial retrieval**.

Paper:
- https://arxiv.org/abs/2510.09894
- https://www.researchgate.net/publication/396458142

*Note: The current preprint provides an early version of the work.  
The complete paper will be released in a forthcoming version.*

AlphaEarth embeddings are introduced by DeepMind in  
https://deepmind.google/blog/alphaearth-foundations-helps-map-our-planet-in-unprecedented-detail/

## Installation

Clone the repository:

git clone <repo-url>
    cd AETHER

Create a Python environment:

python3 -m venv .venv
    source .venv/bin/activate

Install dependencies:

pip install -r requirements.txt

---

## Data Preparation

## AlphaEarth Embeddings

Download AlphaEarth embeddings and place them in:

data/AE/AlphaEarth_embedding.tif

The raster should contain **64-dimensional AlphaEarth embeddings**.

---

## POI Dataset

Prepare a POI dataset containing:

| Field | Description |
|------|-------------|
| id | POI id |
| description | textual description used for embedding|
| geometry | point geometry |

Example file:

data/poi/poi_example.geojson
---

## Text Embeddings

Two text encoding options are supported.

### OpenAI

Set API key:

export OPENAI_API_KEY=your_key_here

Model example:

text-embedding-3-large
    text-embedding-3-small

### HuggingFace

Example model:

sentence-transformers/all-MiniLM-L6-v2

## Training AETHER

Training is controlled by a configuration file.

Run:

python train.py --cfg config.yaml

The training pipeline performs the following steps:

1. Load POI data with textual descriptions and spatial coordinates  
2. Extract AlphaEarth embeddings around POI locations using **spatial window pooling**  
3. Encode POI text using a pretrained language model  
4. Project AE and POI embeddings into a shared latent space  
5. Train a multimodal contrastive alignment model

For each POI location, two spatial views are constructed:

- **Base view:** AE embeddings pooled from a smaller spatial window  
- **Augmented view:** AE embeddings pooled from a larger spatial window

These two views capture spatial context at different scales.

The model jointly optimizes two contrastive objectives:

L = λ L_AE-AE + (1-λ) L_AE-POI

where:

- **L_AE-AE** enforces multi-scale consistency between AE embeddings
- **L_AE-POI** aligns AE embeddings with POI semantic representations

## Inference AETHER embeddings

To transform AlphaEarth embeddings into AETHER embeddings:

```bash
    python infer_aether_raster.py \
    --ae_tif ./data/AE/Alphaearth_embedding_2024.tif \
    --ckpt ./ckpts/epoch_0100.pth \
    --out ./outputs/AETHER_embedding.tif
```

## Citation

If you use AETHER in your research, please cite:

@article{liu2025beyond,
    title   = {Beyond AlphaEarth: Toward Human-Centered Geospatial Foundation Models via POI-Guided Contrastive Learning},
    author  = {Liu, Junyuan and Qin, Quan and Dong, Guangsheng and Wang, Xinglei and Feng, Jiazhuang and Zeng, Zichao and Cheng, Tao},
    journal = {arXiv preprint arXiv:2510.09894},
    year    = {2025},
    url     = {https://arxiv.org/abs/2510.09894}
    }

AETHER

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README