Large Language Models (LLMs) often produce fluent but factually incorrect responses, a phenomenon known as hallucination. Abstention, where the model chooses not to answer and instead outputs phrases such as "I don't know", is a common safeguard. However, existing abstention methods typically rely on post-generation signals, such as generation variations or feedback, which limits their ability to prevent unreliable responses in advance. In this paper, we introduce Aspect-Based Causal Abstention (ABCA), a new framework that enables early abstention by analysing the internal diversity of LLM knowledge through causal inference. This diversity reflects the multifaceted nature of parametric knowledge acquired from various sources, representing diverse aspects such as disciplines, legal contexts, or temporal frames. ABCA estimates causal effects conditioned on these aspects to assess the reliability of knowledge relevant to a given query. Based on these estimates, we enable two types of abstention: Type-1, where aspect effects are inconsistent (knowledge conflict), and Type-2, where aspect effects consistently support abstention (knowledge insufficiency). Experiments on standard benchmarks demonstrate that ABCA improves abstention reliability, achieves state-of-the-art performance, and enhances the interpretability of abstention decisions.

Millions of people cannot eat independently due to a disability, and caregiver-assisted meals can make them feel self-conscious, pressured, or burdensome. Robot-assisted feeding promises to empower people with motor impairments to feed themselves. However, current research typically examines specific robotic system subcomponents and evaluates them in controlled lab settings. This leaves a gap in developing and evaluating an end-to-end system that can feed entire meals in out-of-lab settings. We present one such system, which we developed collaboratively with two community researchers (CRs) with motor-impairments. The key challenge of developing a robot feeding system for out-of-lab use is the varied off-nominal scenarios that inevitably arise. Our key insight is that users can overcome many off-nominals, provided customizability and control over the system. Our system improves upon the state-of-the-art with: (1) a user interface that provides substantial user customizability and control, (2) a bite selection implementation that incorporates users-in-the-loop to generalize across food items, and (3) portable hardware that facilitates system use in diverse environments without inhibiting user mobility. We conduct two studies to evaluate the system. In Study 1, five users with motor impairments and one CR use the system to feed themselves meals of their choice in a cafeteria, office, or conference room. In Study 2, one CR uses the system in his home for five days, feeding himself 10 meals across diverse contexts. We present 3 key lesson learned: (1) spatial contexts are numerous, customizability lets users adapt to them; (2) off-nominals will arise, variable autonomy lets users overcome them; and (3) assistive robots' benefits depend on context. We provide video footage and code on our website.

The increasing frequency of suicidal thoughts highlights the importance of early detection and intervention. Social media platforms, where users often share personal experiences and seek help, could be utilized to identify individuals at risk. However, the large volume of daily posts makes manual review impractical. This paper explores the use of Large Language Models (LLMs) to automatically detect suicidal content in text-based social media posts. We propose a novel method for generating pseudo-labels for unlabeled data by prompting LLMs, along with traditional classification fine-tuning techniques to enhance label accuracy. To create a strong suicide detection model, we develop an ensemble approach involving prompting with Qwen2-72B-Instruct, and using fine-tuned models such as Llama3-8B, Llama3.1-8B, and Gemma2-9B. We evaluate our approach on the dataset of the Suicide Ideation Detection on Social Media Challenge, a track of the IEEE Big Data 2024 Big Data Cup. Additionally, we conduct a comprehensive analysis to assess the impact of different models and fine-tuning strategies on detection performance. Experimental results show that the ensemble model significantly improves the detection accuracy, by 5% points compared with the individual models. It achieves a weight F1 score of 0.770 on the public test set, and 0.731 on the private test set, providing a promising solution for identifying suicidal content in social media. Our analysis shows that the choice of LLMs affects the prompting performance, with larger models providing better accuracy. Our code and checkpoints are publicly available at https://github.com/khanhvynguyen/Suicide_Detection_LLMs.

Teleoperation of mobile manipulators within a home environment can significantly enhance the independence of individuals with severe motor impairments, allowing them to regain the ability to perform self-care and household tasks. There is a critical need for novel teleoperation interfaces to offer effective alternatives for individuals with impairments who may encounter challenges in using existing interfaces due to physical limitations. In this work, we iterate on one such interface, HAT (Head-Worn Assistive Teleoperation), an inertial-based wearable integrated into any head-worn garment. We evaluate HAT through a 7-day in-home study with Henry Evans, a non-speaking individual with quadriplegia who has participated extensively in assistive robotics studies. We additionally evaluate HAT with a proposed shared control method for mobile manipulators termed Driver Assistance and demonstrate how the interface generalizes to other physical devices and contexts. Our results show that HAT is a strong teleoperation interface across key metrics including efficiency, errors, learning curve, and workload. Code and videos are located on our project website.

MOTIVATION: Predicting the binding between T-cell receptor (TCR) and peptide presented by human leucocyte antigen molecule is a highly challenging task and a key bottleneck in the development of immunotherapy. Existing prediction tools, despite exhibiting good performance on the datasets they were built with, suffer from low true positive rates when used to predict epitopes capable of eliciting T-cell responses in patients. Therefore, an improved tool for TCR-peptide prediction built upon a large dataset combining existing publicly available data is still needed. RESULTS: We collected data from five public databases (IEDB, TBAdb, VDJdb, McPAS-TCR, and 10X) to form a dataset of >3 million TCR-peptide pairs, 3.27% of which were binding interactions. We proposed epiTCR, a Random Forest-based method dedicated to predicting the TCR-peptide interactions. epiTCR used simple input of TCR CDR3β sequences and antigen sequences, which are encoded by flattened BLOSUM62. epiTCR performed with area under the curve (0.98) and higher sensitivity (0.94) than other existing tools (NetTCR, Imrex, ATM-TCR, and pMTnet), while maintaining comparable prediction specificity (0.9). We identified seven epitopes that contributed to 98.67% of false positives predicted by epiTCR and exerted similar effects on other tools. We also demonstrated a considerable influence of peptide sequences on prediction, highlighting the need for more diverse peptides in a more balanced dataset. In conclusion, epiTCR is among the most well-performing tools, thanks to the use of combined data from public sources and its use will contribute to the quest in identifying neoantigens for precision cancer immunotherapy. AVAILABILITY AND IMPLEMENTATION: epiTCR is available on GitHub (https://github.com/ddiem-ri-4D/epiTCR).

Polarization information of light in a scene is valuable for various image processing and computer vision tasks. A division-of-focal-plane polarimeter is a promising approach to capture the polarization images of different orientations in one shot, while it requires color-polarization demosaicking. In this paper, we propose a two-step color-polarization demosaicking network (TCPDNet), which consists of two sub-tasks of color demosaicking and polarization demosaicking. We also introduce a reconstruction loss in the YCbCr color space to improve the performance of TCPDNet. Experimental comparisons demonstrate that TCPDNet outperforms existing methods in terms of the image quality of polarization images and the accuracy of Stokes parameters.

We present RCareWorld, a human-centric simulation world for physical and social robotic caregiving designed with inputs from stakeholders. RCareWorld has realistic human models of care recipients with mobility limitations and caregivers, home environments with multiple levels of accessibility and assistive devices, and robots commonly used for caregiving. It interfaces with various physics engines to model diverse material types necessary for simulating caregiving scenarios, and provides the capability to plan, control, and learn both human and robot control policies by integrating with state-of-the-art external planning and learning libraries, and VR devices. We propose a set of realistic caregiving tasks in RCareWorld as a benchmark for physical robotic caregiving and provide baseline control policies for them. We illustrate the high-fidelity simulation capabilities of RCareWorld by demonstrating the execution of a policy learnt in simulation for one of these tasks on a real-world setup. Additionally, we perform a real-world social robotic caregiving experiment using behaviors modeled in RCareWorld. Robotic caregiving, though potentially impactful towards enhancing the quality of life of care recipients and caregivers, is a field with many barriers to entry due to its interdisciplinary facets. RCareWorld takes the first step towards building a realistic simulation world for robotic caregiving that would enable researchers worldwide to contribute to this impactful field. Demo videos and supplementary materials can be found at: https://emprise.cs.cornell.edu/rcareworld/.

BACKGROUND: Automatic cell type identification is essential to alleviate a key bottleneck in scRNA-seq data analysis. While most existing classification tools show good sensitivity and specificity, they often fail to adequately not-classify cells that are missing in the used reference. Additionally, many tools do not scale to the continuously increasing size of current scRNA-seq datasets. Therefore, additional tools are needed to solve these challenges. RESULTS: scAnnotatR is a novel R package that provides a complete framework to classify cells in scRNA-seq datasets using pre-trained classifiers. It supports both Seurat and Bioconductor's SingleCellExperiment and is thereby compatible with the vast majority of R-based analysis workflows. scAnnotatR uses hierarchically organised SVMs to distinguish a specific cell type versus all others. It shows comparable or even superior accuracy, sensitivity and specificity compared to existing tools while being able to not-classify unknown cell types. Moreover, scAnnotatR is the only of the best performing tools able to process datasets containing more than 600,000 cells. CONCLUSIONS: scAnnotatR is freely available on GitHub ( https://github.com/grisslab/scAnnotatR ) and through Bioconductor (from version 3.14). It is consistently among the best performing tools in terms of classification accuracy while scaling to the largest datasets.