Black-Box Batch Active Learning for Regression

Andreas Kirsch

RECORD

Black-Box Batch Active Learning for Regression

Andreas Kirsch

Machine Learning · Submitted 17 Feb 2023

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

The authors' main claim We demonstrate the effectiveness of our approach through extensive experimental evaluations on regression datasets, achieving surprisingly strong performance compared to white-box approaches for deep learning models.

how it is stated · asserted 1.00 read the record →

Comment (0) Read on arXiv ↗

Cite ▾

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, 6,179 input tokens.

What the abstract itself states, as probabilities: claims first 0.07 · states limitation 0.11 · human participants 0.03 · controlled comparison 0.76 · randomized assignment 0.01 · states quantitative result 0.05 · compares with prior results 0.95 · mentions code or data availability 0.02.

The abstract, sentence by sentence

arXiv listing abstract v2. Results and the main claim are underlined; show every role.

Batch active learning is a popular approach for efficiently training machine learning models on large, initially unlabelled datasets by repeatedly acquiring labels for batches of data points. However, many recent batch active learning methods are white-box approaches and are often limited to differentiable parametric models: they score unlabeled points using acquisition functions based on model embeddings or first- and second-order derivatives. In this paper, we propose black-box batch active learning for regression tasks as an extension of white-box approaches. Crucially, our method only relies on model predictions. This approach is compatible with a wide range of machine learning models, including regular and Bayesian deep learning models and non-differentiable models such as random forests. It is rooted in Bayesian principles and utilizes recent kernel-based approaches. This allows us to extend a wide range of existing state-of-the-art white-box batch active learning methods (BADGE, BAIT, LCMD) to black-box models. We demonstrate the effectiveness of our approach through extensive experimental evaluations on regression datasets, achieving surprisingly strong performance compared to white-box approaches for deep learning models.

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	Batch active learning is a popular approach for efficiently training machine learning models on large, initially unlabelled datasets by repeatedly acquiring labels for batches of data points.	background 1.00	0.04	0.02	asserted 1.00
s1	However, many recent batch active learning methods are white-box approaches and are often limited to differentiable parametric models: they score unlabeled points using acquisition functions based on model embeddings or first- and second-order derivatives.	gap 0.68	0.04	0.03	asserted 0.93
s2	In this paper, we propose black-box batch active learning for regression tasks as an extension of white-box approaches.	contribution 1.00	0.81	0.08	asserted 1.00
s3	Crucially, our method only relies on model predictions.	method 0.71	0.77	0.11	asserted 1.00
s4	This approach is compatible with a wide range of machine learning models, including regular and Bayesian deep learning models and non-differentiable models such as random forests.	method 0.51	0.45	0.07	asserted 1.00
s5	It is rooted in Bayesian principles and utilizes recent kernel-based approaches.	method 0.98	0.47	0.04	asserted 1.00
s6	This allows us to extend a wide range of existing state-of-the-art white-box batch active learning methods (BADGE, BAIT, LCMD) to black-box models.	contribution 0.76	0.80	0.10	asserted 0.97
s7	We demonstrate the effectiveness of our approach through extensive experimental evaluations on regression datasets, achieving surprisingly strong performance compared to white-box approaches for deep learning models.	result 0.99	0.96	0.78	asserted 1.00

What kind of paper the classifier reports this is: method or tool 1.00 · evidence from observation or experiment 0.96.

What was examined, and what was not

sources read

arXiv listing abstract v2

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

Batch active learning is a popular approach for efficiently training machine learning models on large, initially unlabelled datasets by repeatedly acquiring labels for batches of data points. However, many recent batch active learning methods are white-box approaches and are often limited to differentiable parametric models: they score unlabeled points using acquisition functions based on model embeddings or first- and second-order derivatives. In this paper, we propose black-box batch active learning for regression tasks as an extension of white-box approaches. Crucially, our method only relies on model predictions. This approach is compatible with a wide range of machine learning models, including regular and Bayesian deep learning models and non-differentiable models such as random forests. It is rooted in Bayesian principles and utilizes recent kernel-based approaches. This allows us to extend a wide range of existing state-of-the-art white-box batch active learning methods (BADGE, BAIT, LCMD) to black-box models. We demonstrate the effectiveness of our approach through extensive experimental evaluations on regression datasets, achieving surprisingly strong performance compared to white-box approaches for deep learning models.

Human review

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

How the literature uses it

1 arXiv paper cite this in 1 sentence

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

Papers citing it

Cites it as support 1 paper

This property has been demonstrated across many machine learning methods [ref] .

(Implicit) Ensembles of Ensembles: Epistemic Uncertainty Collapse in Large Models · Theoretical Framework · 0.98

What it cites · 17 sentences without a confident label

Builds on 2 papers

We extend the work of [ref] and [ref] by combining the prediction-based approach with a kernel-based formulation.

Bayesian Active Learning for Classification and Preference Learning · Related Work · 1.00
We extend the work of [ref] and [ref] by combining the prediction-based approach with a kernel-based formulation.

A Framework and Benchmark for Deep Batch Active Learning for Regression · Related Work · 1.00

Uses the method or tool of 3 papers

We use the same experimental setup and hyperparameters as [ref] .

A Framework and Benchmark for Deep Batch Active Learning for Regression · Results · 0.89
For non-differentiable models, we experiment with random forests [ref] , using the different trees as ensemble members, and a virtual ensemble of gradient-boosted decision trees [ref] .

CatBoost: unbiased boosting with categorical features · Results · 1.00
We use the implementation in CatBoost [ref] .

CatBoost: gradient boosting with categorical features support · Results · 1.00

Compares with 4 papers

Specifically, we compare to the following popular deep active learning methods: BALD [ref] , BatchBALD [ref] , BAIT [ref] , BADGE [ref] , ACS-FW [ref] , Core-Set [ref] /FF-Active [ref] , and LCMD [ref] .

Active Learning for Convolutional Neural Networks: A Core-Set Approach · Results · 0.99
Specifically, we compare to the following popular deep active learning methods: BALD [ref] , BatchBALD [ref] , BAIT [ref] , BADGE [ref] , ACS-FW [ref] , Core-Set [ref] /FF-Active [ref] , and LCMD [ref] .

Deep Active Learning over the Long Tail · Results · 0.99
Specifically, we compare to the following popular deep active learning methods: BALD [ref] , BatchBALD [ref] , BAIT [ref] , BADGE [ref] , ACS-FW [ref] , Core-Set [ref] /FF-Active [ref] , and LCMD [ref] .

A Framework and Benchmark for Deep Batch Active Learning for Regression · Results · 0.99

Cites as background 2 papers

For other selection methods, we refer to [ref] .

A Framework and Benchmark for Deep Batch Active Learning for Regression · Kernel-based Methods & Information Theory · 0.98
A comprehensive survey of early active learning methods can be found in [ref] , while more recent surveys of contemporary deep learning methods can be found in [ref] and [ref] .

A Comparative Survey of Deep Active Learning · Related Work · 1.00

Discussion

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