Poster Location: Halle B #345

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.

Poster Location: Halle B #346

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.

Poster Location: Halle B #347

Multi-intent learning with limited supervision involves predicting multiple intentions of utterances using only a few annotated samples. The primary motivation for this task stems from the high costs and cumbersome processes associated with annotating large datasets. To mitigate this, we propose utilising Large Language Models (LLMs) for annotation assistance. Although LLMs show promise, they struggle with response randomness, and their previous prompts is static and do not learn from their outputs. To address this, we propose `self-teaching prompting' (STP), a method that enables Large Language Models (LLMs) to iteratively learn from their consistent samples and refine their predictions over time. Our experiments with multi-intention datasets demonstrate that STP significantly enhances response accuracy.

Poster Location: Halle B #348

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

Poster Location: Halle B #349

Natural Adversarial Examples (NAEs), images arising naturally from the environment and capable of deceiving classifiers, are instrumental in robustly evaluating and identifying vulnerabilities in trained models. In this work, unlike prior works that passively collect NAEs from real images, we propose to actively synthesize NAEs using the state-of-the-art Stable Diffusion. Specifically, our method formulates a controlled optimization process, where we perturb the token embedding that corresponds to a specified class to generate NAEs. This generation process is guided by the gradient of loss from the target classifier, ensuring that the created image closely mimics the ground-truth class yet fools the classifier. Named SD-NAE (Stable Diffusion for Natural Adversarial Examples), our innovative method is effective in producing valid and useful NAEs, which is demonstrated through a meticulously designed experiment. Code is available at https://anonymous.4open.science/r/SD-NAE/.

Poster Location: Halle B #350

Mixture density networks (MDNs) are a natural choice to model multi-modal predictions for trajectory prediction or motion planning.However, MDNs are often difficult to train due to mode collapse and a need for careful initialization, which becomes even more problematic when the number of mixture components are strongly overestimated. To address this issue in motion planning problems, we propose a pre-training scheme for MDNs called common sense initialization (CSI). Pre-training with CSI allows variety-encouraging optimization such as Winner-Takes-All (WTA) to exploit the initialized weights during training so that the MDN can converge when the number of components are overestimated. This paper presents empirical evidence for the effectiveness of CSI when applied to motion planning of pedestrian agents in urban environments.