Affinity Workshop
Tiny Papers Oral Session 1
Thomas F Burns · Krystal Maughan
Halle A 8 - 9
Live content is unavailable. Log in and register to view live content
Schedule
-
|
TOWARDS FAIRNESS CONSTRAINED RESTLESS MULTI-ARMED BANDITS: A CASE STUDY OF MATERNAL AND CHILD CARE DOMAIN
(
Oral
)
>
link
Restless multi-armed bandits (RMABs) are widely used for resource allocation in dynamic environments, but they typically do not consider fairness implications. This paper introduces a fairness-aware approach for offline RMABs. We propose a Kullback-Leibler (KL) divergence-based fairness metric to quantify the discrepancy between the selected and the overall population. This is incorporated as a regularizer into the soft whittle index optimization. We evaluate our fairness-aware algorithm on a real-world RMAB dataset where initial results suggest that our approach can potentially improve fairness while preserving solution quality. |
Gargi Singh · Milind Tambe · Aparna Taneja 🔗 |
-
|
Transfer Learning for Global Feature Importance Measurements
(
Oral
)
>
link
Understanding feature importance is crucial for conducting interpretable clinical decision-making. However, the reliability of such analyses can be heavily impacted by the available sample size, placing sites with lower data quality and smaller sample sizes at inherent disadvantages. To address the challenge, we propose a model-agnostic transfer learning-based approach for feature importance measurement and evaluate its effectiveness using real-world heterogeneous electronic health records. |
Xin Li · Siqi Li · Qiming Wu · Kunyu Yu 🔗 |
-
|
Aligners: Decoupling LLMs and Alignment
(
Oral
)
>
link
Large Language Models (LLMs) need to be aligned with human expectations to ensure their safety and utility in most applications. Alignment is challenging, costly, and needs to be repeated for every LLM and alignment criterion. We propose to decouple LLMs and alignment by training aligner models that can be used to align any LLM for a given criteria on an as-needed basis, thus also reducing the potential negative impacts of alignment on performance. Our recipe for training the aligner models solely relies on synthetic data generated with a (prompted) LLM and can be easily adjusted for a variety of alignment criteria. We illustrate our method by training an ``ethical'' aligner and verify its efficacy empirically. |
Lilian Ngweta · Mayank Agarwal · Subha Maity · Alex Gittens · Yuekai Sun · Mikhail Yurochkin 🔗 |
-
|
VISUAL PROMPTING METHODS FOR GPT-4V BASED ZERO-SHOT GRAPHIC LAYOUT DESIGN GENERATION
(
Oral
)
>
link
Graphic layout design generation is a challenging problem in computer vision. The key aspect of the challenge is ensuring coherent placement of textual elements on the background image to ensure aesthetic appeal and avoiding occlusion of key visual elements. Although prior methods have made attempts to solve this multi-modal problem, they couldn't perfect it. Owing to the complexity required in understanding the relationship between visual and text elements in the aforementioned task, we investigate GPT-4-Vision(GPT-4V), a large multimodal models(LMMs), to do zero-shot graphic layout design generation in a versatile manner. Our approach explores various off-the-shelf segmentation/superpixel methods to identify and mark the key regions to visually augment the image to enhance GPT-4V's spatial reasoning capability . The results of our comprehensive experiments on a self-curated dataset demonstrates the efficacy of our proposed visual prompting methods, showing improvement over standard GPT-4V prompting method and also performing at par and even better, for some techniques, than state-of-the-art specialist model.The code and data is available at https://anonymous.4open.science/r/VISUAL-PROMPTING-TECHNIQUES-FOR-GPT-4V-BASED-ZERO-SHOT-GRAPHIC-LAYOUT-DESIGN-GENERATION-5A6E |
Kunal Singh · Mukund Khanna · Ankan Biswas · Pradeep Moturi · Shivam 🔗 |
-
|
Chemical Language Models Have Problems with Chemistry: A Case Study on Molecule Captioning Task
(
Oral
)
>
link
Drug discovery has been greatly enhanced through the recent fusion of molecular sciences and natural language processing, leading these research fields to significant advancements. Considering the crucial role of molecule representation in chemical understanding within these models, we introduce novel probing tests designed to evaluate chemical knowledge of molecular structure in state-of-the-art language models (LMs), specifically MolT5 and Chem+Text T5. These probing tests are conducted on a molecule captioning task to gather evidence and insights into the language models' comprehension of chemical information. By applying rules to transform molecular SMILES into equivalent variants, we have observed significant differences in the natural language descriptions generated by the LM for a given molecule depending on the exact transformation used. |
Kuzma Khrabrov · Elena Tutubalina 🔗 |
-
|
Is Watermarking LLM-Generated Code Robust?
(
Oral
)
>
link
We present the first study of the robustness of existing watermarking techniques on Python code generated by large language models. Although existing works showed that watermarking can be robust for natural language text, we show that it is easy to remove these watermarks on code by simple semantic-preserving transformations. |
Tarun Suresh 🔗 |
-
|
Rescaling Intermediate Features Makes Trained Consistency Models Perform Better
(
Oral
)
>
link
In the domain of deep generative models, diffusion models are renowned for their high-quality image generation but are constrained by intensive computational demands. To mitigate this, consistency models have been proposed as a computationally efficient alternative. Our research reveals that post-training rescaling of internal features can enhance the one-step sample quality of these models without incurring detectable computational overhead. This optimization is evidenced by an obvious improvement in Fréchet Inception Distance (FID). For example, with our rescaled consistency distillation (CD) model, FID on the ImageNet dataset reduces from 6.2 to 5.2, on the LSUN-cat dataset from 10.9 to 9.5. Closer inspection of the generated images reveals that this enhancement may originate from improved visual details and clarity. |
Junyi Zhu 🔗 |