Neural Embedding Compression For Efficient Multi-Task Earth Observation Modelling

Carlos Gomes; Thomas Brunschwiler

RECORD

Neural Embedding Compression For Efficient Multi-Task Earth Observation Modelling

Carlos Gomes, Thomas Brunschwiler

Machine Learning · Submitted 26 Mar 2024

abstract compiled · 2026-09-23 · no language-model review

The authors' main claim Overall, NEC is a data-efficient yet performant approach for multi-task EO modelling.

how it is stated · asserted 1.00 read the record →

Comment (0) Read on arXiv ↗

Cite ▾

arXiv ID2403.17886 (version 5)

Pith IDAYKZ62PJ, Pith's permanent record of this paper and its review

CitationCarlos Gomes, Thomas Brunschwiler. "Neural Embedding Compression For Efficient Multi-Task Earth Observation Modelling." arXiv:2403.17886 (2024). https://pith.science/paper/AYKZ62PJ

Claim paper

The compiled record

The paper's own sentences, arranged. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.

Open the compiled record

Compiled by classifier-1.13.0 via TypeSafe API, question bank abstract-p1-20260923, 2026-09-23: 1 request, 6,126 input tokens.

What the abstract itself states, as probabilities: claims first 0.05 · states limitation 0.59 · human participants 0.02 · controlled comparison 0.90 · randomized assignment 0.02 · states quantitative result 0.99 · compares with prior results 0.92 · mentions code or data availability 0.03.

The abstract, sentence by sentence

arXiv listing abstract v5. Results and the main claim are underlined; show every role.

As repositories of large scale data in earth observation (EO) have grown, so have transfer and storage costs for model training and inference, expending significant resources. We introduce Neural Embedding Compression (NEC), based on the transfer of compressed embeddings to data consumers instead of raw data. We adapt foundation models (FM) through learned neural compression to generate multi-task embeddings while navigating the tradeoff between compression rate and embedding utility. We update only a small fraction of the FM parameters (10%) for a short training period (1% of the iterations of pre-training). We evaluate NEC on two EO tasks: scene classification and semantic segmentation. Compared with applying traditional compression to the raw data, NEC achieves similar accuracy with a 75% to 90% reduction in data. Even at 99.7% compression, performance drops by only 5% on the scene classification task. Overall, NEC is a data-efficient yet performant approach for multi-task EO modelling.

background contribution method result interpretation main claim

What the classifier reports for each sentence. The number is the probability it gave the label.
#	sentence	role	the authors' own work	states the main conclusion	strength
s0	As repositories of large scale data in earth observation (EO) have grown, so have transfer and storage costs for model training and inference, expending significant resources.	background 1.00	0.07	0.03	asserted 1.00
s1	We introduce Neural Embedding Compression (NEC), based on the transfer of compressed embeddings to data consumers instead of raw data.	contribution 0.96	0.87	0.08	asserted 0.99
s2	We adapt foundation models (FM) through learned neural compression to generate multi-task embeddings while navigating the tradeoff between compression rate and embedding utility.	method 0.91	0.91	0.07	asserted 0.99
s3	We update only a small fraction of the FM parameters (10%) for a short training period (1% of the iterations of pre-training).	method 0.98	0.94	0.03	asserted 1.00
s4	We evaluate NEC on two EO tasks: scene classification and semantic segmentation.	method 0.62	0.95	0.04	asserted 0.98
s5	Compared with applying traditional compression to the raw data, NEC achieves similar accuracy with a 75% to 90% reduction in data.	result 1.00	0.95	0.25	asserted 1.00
s6	Even at 99.7% compression, performance drops by only 5% on the scene classification task.	result 1.00	0.95	0.10	asserted 1.00
s7	Overall, NEC is a data-efficient yet performant approach for multi-task EO modelling.	interpretation 0.97	0.71	0.92	asserted 1.00

What kind of paper the classifier reports this is: method or tool 1.00 · evidence from observation or experiment 0.84.

What was examined, and what was not

sources read

arXiv listing abstract v5

not examined

full text
figures and tables
references and citing sentences
relations to other papers

levels run

P abstract

not run

S sections
T body sentences
C citations
R cross-paper relations

Every quoted sentence on this record is the authors'. Every label is from Pith's closed vocabulary, with the probability shown beside it. Every number in the measurements table is the authors', checked by code. Record as JSON

The paper

Original abstract

As repositories of large scale data in earth observation (EO) have grown, so have transfer and storage costs for model training and inference, expending significant resources. We introduce Neural Embedding Compression (NEC), based on the transfer of compressed embeddings to data consumers instead of raw data. We adapt foundation models (FM) through learned neural compression to generate multi-task embeddings while navigating the tradeoff between compression rate and embedding utility. We update only a small fraction of the FM parameters (10%) for a short training period (1% of the iterations of pre-training). We evaluate NEC on two EO tasks: scene classification and semantic segmentation. Compared with applying traditional compression to the raw data, NEC achieves similar accuracy with a 75% to 90% reduction in data. Even at 99.7% compression, performance drops by only 5% on the scene classification task. Overall, NEC is a data-efficient yet performant approach for multi-task EO modelling.

Human review

How Pith checks papers → · Full audit of this paper →

How the literature uses it

What this paper does with the arXiv papers it cites

Each citing sentence is taken from the citing paper's LaTeX source and typed by Pith's classifier. A label appears at probability 0.8 or more (0.95 for support), a contest at 0.7 or more; the quoted words are the citing authors'. How each label was checked · JSON

What it cites · 4 sentences without a confident label

Uses the method or tool of 3 papers

In this section, we frame our method as the optimization of a rate-distortion objective following the framework of end-to-end learned compression established in Balle et al. [ref] .

End-to-end Optimized Image Compression · METHOD · 0.96
Specifically, we leverage the implementation in the CompressAI [ref] library.

CompressAI: a PyTorch library and evaluation platform for end-to-end compression research · METHOD · 1.00
We process it as is standard [ref] , obtaining tiles of size $512 ×512$.

AerialFormer: Multi-resolution Transformer for Aerial Image Segmentation · EXPERIMENTS · 0.87

Cites as background 9 papers

In this study, we propose the application of learned neural compression, which has been well studied in the domain of image and video compression [ref] .

DVC: An End-to-end Deep Video Compression Framework · INTRODUCTION · 0.85
In this study, we propose the application of learned neural compression, which has been well studied in the domain of image and video compression [ref] .

End-to-end Optimized Image Compression · INTRODUCTION · 0.85
Recently, some works have leveraged masked auto-encoding (MAE) as a lossy compression method, dropping data at the image patch [ref] or embedding level [ref] .

Exploring Effective Mask Sampling Modeling for Neural Image Compression · RELATED WORK · 0.93

Discussion

Know something about this paper? Leave a comment, or highlight any passage above to comment on it.