Skip to content

Commit 5ae0303

Browse files
lawrenndlawrennd
committed
Add pages for volume v260
0 parents  commit 5ae0303

File tree

92 files changed

+7508
-0
lines changed

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

92 files changed

+7508
-0
lines changed

‎Gemfile

Lines changed: 15 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,15 @@
1+
source "https://rubygems.org"
2+
3+
git_source(:github) {|repo_name| "https://github.com/#{repo_name}" }
4+
5+
gem 'jekyll'
6+
7+
group :jekyll_plugins do
8+
gem 'github-pages'
9+
gem 'jekyll-remote-theme'
10+
gem 'jekyll-include-cache'
11+
gem 'webrick'
12+
end
13+
14+
# gem "rails"
15+

‎README.md

Lines changed: 22 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,22 @@
1+
# PMLR 260
2+
3+
To suggest fixes to this volume please make a pull request containing the changes requested and a justification for the changes.
4+
5+
To edit the details of this conference work edit the [_config.yml](./_config.yml) file and submit a pull request.
6+
7+
To make changes to the individual paper details, edit the associated paper file in the [./_posts](./_posts) subdirectory.
8+
9+
For details of how to publish in PMLR please check https://proceedings.mlr.press/faq.html
10+
11+
For details of what is required to submit a proceedings please check https://proceedings.mlr.press/spec.html
12+
13+
14+
15+
Published as Volume 260 by the Proceedings of Machine Learning Research on 14 January 2025.
16+
17+
Volume Edited by:
18+
* Vu Nguyen
19+
* Hsuan-Tien Lin
20+
21+
Series Editors:
22+
* Neil D. Lawrence

‎_config.yml

Lines changed: 89 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,89 @@
1+
---
2+
booktitle: Proceedings of the 16th Asian Conference on Machine Learning
3+
shortname: ACML
4+
publisher: PMLR
5+
published: 2025-01-14
6+
start: &1 2024-12-05
7+
end: 2024-12-08
8+
series: Proceedings of Machine Learning Research
9+
volume: '260'
10+
year: '2025'
11+
layout: proceedings
12+
issn: 2640-3498
13+
id: ACML2024
14+
month: 0
15+
cycles: false
16+
bibtex_editor: Nguyen, Vu and Lin, Hsuan-Tien
17+
editor:
18+
- given: Vu
19+
family: Nguyen
20+
- given: Hsuan-Tien
21+
family: Lin
22+
title: Proceedings of Machine Learning Research
23+
description: |
24+
Proceedings of the 16th Asian Conference on Machine Learning
25+
Held in Hanoi, Vietnam on 05-08 December 2024
26+
27+
Published as Volume 260 by the Proceedings of Machine Learning Research on 14 January 2025.
28+
29+
Volume Edited by:
30+
Vu Nguyen
31+
Hsuan-Tien Lin
32+
33+
Series Editors:
34+
Neil D. Lawrence
35+
date_str: 05--08 Dec
36+
url: https://proceedings.mlr.press
37+
author:
38+
name: PMLR
39+
baseurl: "/v260"
40+
twitter_username: MLResearchPress
41+
github_username: mlresearch
42+
markdown: kramdown
43+
exclude:
44+
- README.md
45+
- Gemfile
46+
- ".gitignore"
47+
plugins:
48+
- jekyll-feed
49+
- jekyll-seo-tag
50+
- jekyll-remote-theme
51+
remote_theme: mlresearch/jekyll-theme
52+
style: pmlr
53+
permalink: "/:title.html"
54+
ghub:
55+
edit: true
56+
repository: v260
57+
display:
58+
copy_button:
59+
bibtex: true
60+
endnote: true
61+
apa: true
62+
comments: false
63+
volume_type: Volume
64+
volume_dir: v260
65+
email: ''
66+
conference:
67+
name: Asian Conference on Machine Learning
68+
url: https://www.acml-conf.org/2024/
69+
location: Hanoi, Vietnam
70+
dates:
71+
- *1
72+
- 2024-12-06
73+
- 2024-12-07
74+
- 2024-12-08
75+
analytics:
76+
google:
77+
tracking_id: UA-92432422-1
78+
link_visibility:
79+
openreview: true
80+
pdf: true
81+
supplementary: true
82+
software: true
83+
video: true
84+
arxiv: true
85+
doi: true
86+
website: true
87+
orig_bibfile: "/Users/neil/mlresearch/v260/acml2024.bib"
88+
# Site settings
89+
# Original source: /Users/neil/mlresearch/v260/acml2024.bib

‎_posts/2025-01-14-aravindan25a.md

Lines changed: 65 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,65 @@
1+
---
2+
title: 'EVaDE : Event-Based Variational Thompson Sampling for Model-Based Reinforcement
3+
Learning'
4+
booktitle: Proceedings of the 16th Asian Conference on Machine Learning
5+
year: '2025'
6+
volume: '260'
7+
series: Proceedings of Machine Learning Research
8+
month: 0
9+
publisher: PMLR
10+
pdf: https://raw.githubusercontent.com/mlresearch/v260/main/assets/aravindan25a/aravindan25a.pdf
11+
url: https://proceedings.mlr.press/v260/aravindan25a.html
12+
software: https://tinyurl.com/3zb8nywx
13+
openreview: OYlNXMMSXB
14+
abstract: Posterior Sampling for Reinforcement Learning (PSRL) is a well-known algorithm
15+
that augments model-based reinforcement learning (MBRL) algorithms with Thompson
16+
sampling. PSRL maintains posterior distributions of the environment transition dynamics
17+
and the reward function, which are intractable for tasks with high-dimensional state
18+
and action spaces. Recent works show that dropout, used in conjunction with neural
19+
networks, induces variational distributions that can approximate these posteriors.
20+
In this paper, we propose Event-based Variational Distributions for Exploration
21+
(EVaDE), which are variational distributions that are useful for MBRL, especially
22+
when the underlying domain is object-based. We leverage the general domain knowledge
23+
of object-based domains to design three types of event-based convolutional layers
24+
to direct exploration. These layers rely on Gaussian dropouts and are inserted between
25+
the layers of the deep neural network model to help facilitate variational Thompson
26+
sampling. We empirically show the effectiveness of EVaDE-equipped Simulated Policy
27+
Learning (EVaDE-SimPLe) on the 100K Atari game suite.
28+
layout: inproceedings
29+
issn: 2640-3498
30+
id: aravindan25a
31+
tex_title: "{EVaDE }: {E}vent-Based Variational Thompson Sampling for Model-Based
32+
Reinforcement Learning"
33+
firstpage: 559
34+
lastpage: 574
35+
page: 559-574
36+
order: 559
37+
cycles: false
38+
bibtex_editor: Nguyen, Vu and Lin, Hsuan-Tien
39+
editor:
40+
- given: Vu
41+
family: Nguyen
42+
- given: Hsuan-Tien
43+
family: Lin
44+
bibtex_author: Aravindan, Siddharth and Mittal, Dixant and Lee, Wee Sun
45+
author:
46+
- given: Siddharth
47+
family: Aravindan
48+
- given: Dixant
49+
family: Mittal
50+
- given: Wee Sun
51+
family: Lee
52+
date: 2025-01-14
53+
address:
54+
container-title: Proceedings of the 16th Asian Conference on Machine Learning
55+
genre: inproceedings
56+
issued:
57+
date-parts:
58+
- 2025
59+
- 1
60+
- 14
61+
extras:
62+
- label: Supplementary PDF
63+
link: https://raw.githubusercontent.com/mlresearch/v260/main/assets/assets/aravindan25a/aravindan25a-supp.pdf
64+
# Format based on Martin Fenner's citeproc: https://blog.front-matter.io/posts/citeproc-yaml-for-bibliographies/
65+
---

‎_posts/2025-01-14-bougie25a.md

Lines changed: 64 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,64 @@
1+
---
2+
title: 'Exploring Beyond Curiosity Rewards: Language-Driven Exploration in RL'
3+
booktitle: Proceedings of the 16th Asian Conference on Machine Learning
4+
year: '2025'
5+
volume: '260'
6+
series: Proceedings of Machine Learning Research
7+
month: 0
8+
publisher: PMLR
9+
pdf: https://raw.githubusercontent.com/mlresearch/v260/main/assets/bougie25a/bougie25a.pdf
10+
url: https://proceedings.mlr.press/v260/bougie25a.html
11+
openreview: qHv7qTETsw
12+
abstract: 'Sparse rewards pose a significant challenge for many reinforcement learning
13+
algorithms, which struggle in the absence of a dense, well-shaped reward function.
14+
Drawing inspiration from the curiosity exhibited in animals, intrinsically-driven
15+
methods overcome this drawback by incentivizing agents to explore novel states.
16+
Yet, in the absence of domain-specific priors, sample efficiency is hindered as
17+
most discovered novelty has little relevance to the true task reward. We present
18+
iLLM, a curiosity-driven approach that leverages the inductive bias of foundation
19+
models — Large Language Models, as a source of information about plausibly useful
20+
behaviors. Two tasks are introduced for shaping exploration: 1) action generation
21+
and 2) history compression, where the language model is prompted with a description
22+
of the state-action trajectory. We further propose a technique for mapping state-action
23+
pairs to pretrained token embeddings of the language model in order to alleviate
24+
the need for explicit textual descriptions of the environment. By distilling prior
25+
knowledge from large language models, iLLM encourages agents to discover diverse
26+
and human-meaningful behaviors without requiring direct human intervention. We evaluate
27+
the proposed method on BabyAI-Text, MiniHack, Atari games, and Crafter tasks, demonstrating
28+
higher sample efficiency compared to prior curiosity-driven approaches.'
29+
layout: inproceedings
30+
issn: 2640-3498
31+
id: bougie25a
32+
tex_title: "{Exploring Beyond Curiosity Rewards}: {L}anguage-Driven Exploration in
33+
{RL}"
34+
firstpage: 127
35+
lastpage: 142
36+
page: 127-142
37+
order: 127
38+
cycles: false
39+
bibtex_editor: Nguyen, Vu and Lin, Hsuan-Tien
40+
editor:
41+
- given: Vu
42+
family: Nguyen
43+
- given: Hsuan-Tien
44+
family: Lin
45+
bibtex_author: Bougie, Nicolas and Watanabe, Narimasa
46+
author:
47+
- given: Nicolas
48+
family: Bougie
49+
- given: Narimasa
50+
family: Watanabe
51+
date: 2025-01-14
52+
address:
53+
container-title: Proceedings of the 16th Asian Conference on Machine Learning
54+
genre: inproceedings
55+
issued:
56+
date-parts:
57+
- 2025
58+
- 1
59+
- 14
60+
extras:
61+
- label: Supplementary PDF
62+
link: https://raw.githubusercontent.com/mlresearch/v260/main/assets/assets/bougie25a/bougie25a-supp.pdf
63+
# Format based on Martin Fenner's citeproc: https://blog.front-matter.io/posts/citeproc-yaml-for-bibliographies/
64+
---

‎_posts/2025-01-14-cai25a.md

Lines changed: 75 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,75 @@
1+
---
2+
title: 'FTP: A Human Pose Estimation Method Integrating Temporal and Fine-Grained
3+
Feature Fusion'
4+
booktitle: Proceedings of the 16th Asian Conference on Machine Learning
5+
year: '2025'
6+
volume: '260'
7+
series: Proceedings of Machine Learning Research
8+
month: 0
9+
publisher: PMLR
10+
pdf: https://raw.githubusercontent.com/mlresearch/v260/main/assets/cai25a/cai25a.pdf
11+
url: https://proceedings.mlr.press/v260/cai25a.html
12+
openreview: 9XSm6cUhvS
13+
abstract: Human pose estimation is a significant research direction in the field of
14+
computer vision, with critical applications in human motion reconstruction and analysis.
15+
Currently proposed human pose estimation methods primarily focus on single-modality
16+
sensor information, such as RGB images and LiDAR point clouds. While these methods
17+
have achieved promising results within their respective domains, they remain limited
18+
by the inherent deficiencies of each modality, hindering their applicability across
19+
diverse real-world scenarios. With the recent introduction of numerous multi-modality
20+
human pose datasets, multi-modality approaches have begun to develop. However, existing
21+
multi-modality fusion methods mainly consider the global feature relationships between
22+
different modalities, without modeling finer-grained features or the dynamic temporal
23+
relationships between modalities. To address this issue, we propose a novel pipeline
24+
that integrates point cloud and image features, explicitly encoding fine-grained
25+
features and dynamic temporal relationships between the two modalities. Additionally,
26+
we employ a discriminator structure for semi-supervised training. Extensive experiments
27+
demonstrate that our method achieves state-of-the-art (SOTA) performance compared
28+
to previous methods.
29+
layout: inproceedings
30+
issn: 2640-3498
31+
id: cai25a
32+
tex_title: "{FTP}: {A} Human Pose Estimation Method Integrating Temporal and Fine-Grained
33+
Feature Fusion"
34+
firstpage: 858
35+
lastpage: 872
36+
page: 858-872
37+
order: 858
38+
cycles: false
39+
bibtex_editor: Nguyen, Vu and Lin, Hsuan-Tien
40+
editor:
41+
- given: Vu
42+
family: Nguyen
43+
- given: Hsuan-Tien
44+
family: Lin
45+
bibtex_author: Cai, Shuqiang and Ma, Chennan and Wang, Xin and Lin, Li and Yan, Ming
46+
and Lin, Xincheng and Fan, Shuqi and Shen, Siqi
47+
author:
48+
- given: Shuqiang
49+
family: Cai
50+
- given: Chennan
51+
family: Ma
52+
- given: Xin
53+
family: Wang
54+
- given: Li
55+
family: Lin
56+
- given: Ming
57+
family: Yan
58+
- given: Xincheng
59+
family: Lin
60+
- given: Shuqi
61+
family: Fan
62+
- given: Siqi
63+
family: Shen
64+
date: 2025-01-14
65+
address:
66+
container-title: Proceedings of the 16th Asian Conference on Machine Learning
67+
genre: inproceedings
68+
issued:
69+
date-parts:
70+
- 2025
71+
- 1
72+
- 14
73+
extras: []
74+
# Format based on Martin Fenner's citeproc: https://blog.front-matter.io/posts/citeproc-yaml-for-bibliographies/
75+
---

0 commit comments

Comments
 (0)