In today’s world, emotional support is increasingly essential, yet it remains challenging for both those seeking help and those offering it. Multimodal approaches to emotional support show great promise by integrating diverse data sources to provide empathetic, contextually relevant responses, fostering more effective interactions. However, current methods have notable limitations, often relying solely on text or converting other data types into text, or providing emotion recognition only, thus overlooking the full potential of multimodal inputs. Moreover, many studies prioritize response generation without accurately identifying critical emotional support elements or ensuring the reliability of outputs. To overcome these issues, we introduce MULTIMOOD, a new framework that (i) leverages multimodal embeddings from video, audio, and text to predict emotional components and to produce responses responses aligned with professional therapeutic standards. To improve trustworthiness, we (ii) incorporate novel psychological criteria and apply Reinforcement Learning (RL) to optimize large language models (LLMs) for consistent adherence to these standards. We also (iii) analyze several advanced LLMs to assess their multimodal emotional support capabilities. Experimental results show that MultiMood achieves state-of-the-art on MESC and DFEW datasets while RL-driven trustworthiness improvements are validated through human and LLM evaluations, demonstrating its superior capability in applying a multimodal framework in this domain.

The definition of words is the fundamental and crucial linguistic concept. Any changes in word definition lead to changes in the theoretical system of the respective language. Traditionally, researchers in Natural Language Processing (NLP) for Vietnamese texts believe Vietnamese words are constructed from syllables. However, their works did not explicitly mention which linguistic theory they followed for this assumption. Although there are no theoretical guarantees, most NLP studies in Vietnamese accept this assumption. Consequently, word segmentation is recognized as one of the essential stages in NLP for Vietnamese texts. In this study, we address the role of word segmentation for Vietnamese texts from linguistic perspectives. Through our extensive experiments, we show that, based on linguistic theories, performing word segmentation is not appropriate for Vietnamese text understanding. Moreover, we present a novel method, Vietnamese Word TransFormer (ViWordFormer), for modeling Vietnamese word formation. Experimental results indicate that our method is appropriate for modeling Vietnamese word formation from both theoretical and experimental aspects and embark on a novel approach to Vietnamese word representation.

If legal professionals are to use Large Language Models (LLMs), LLMs must provide explanations to empower safe decision making. Argumentative LLMs (ArgLLMs) were recently introduced to produce and explain decisions in the form of claim verifications; they provide explanations that faithfully match their underlying reasoning and are contestable by human users. In this paper, ArgLLMs are applied to two COLIEE 2025 legal entailment tasks: task 4, which involves determining whether relevant statute articles entail a legal hypothesis, and the pilot task, which involves determining whether tort case information entails the conclusion that the tort case was affirmed by the judge. We perform an ablation study to assess how several modifications to ArgLLMs affect accuracy in the two tasks. We also compare the performance of our variants of ArgLLMs against two baselines, one involving a single zero-shot Chain of Thought (CoT) prompt and another involving a pairwise comparison of supporting and attacking arguments. Our experiments show that ArgLLMs are more accurate than over half of all official submissions to task 4 and the pilot task of the COLIEE competition. Moreover, Arg-LLMs produce accuracy scores similar to the CoT baseline, while also providing the benefit of faithful and contestable explanations behind the decision made. For task 4, we also conducted a pilot study where legal experts reviewed 20 generated explanations and found the arguments on the correct side of the debate to be both sound and faithful to the legal context provided in 17 of them. Repository link: https://github.com/charlieblindsay/arg-llms

This paper presents a flexible dual-type proximity sensor forrobotic applications such as human collaborative robots(HCRs) todetect the surroundings. The sensor consists of two parts; sensing transducer and a shielding layer. To amplify the sensing performance, a resonant frequency is formed by an inductive(L-type) electrode and two capacitive(C-type) electrodes, which are placed in coplanar with an LCR circuit. An optimal frequency range is suggested to amplify the proximity detecting performance with a consistent response of impedance change. The developed sensor has a size of 100 x120x2.8 mm3. To obtain the flexibility for various robot surfaces attachment, the electrode layer is made of Flexible Printed (a) Circuit Board(FPCB). Combined with the grounded shielding layer, the sensor can detect objects up to 300 mm in 10 mm resolution when attached to a grounded surface. The sensor is evaluated in diverse circumstances to validity for practical use on robotics.

This paper introduces a new hole-filling method using extrapolated spatio-temporal background information to obtain a synthesized free-view. New temporal background modeling is proposed, which incorporates stationary temporal information in the hole-filling process and preserves the temporal consistency of synthesized views. A background codebook is distinguished from a non-overlapped patch-based codebook, which contributes to extracting reliable temporal background information. Furthermore, a depth-map driven spatial local background estimation is also addressed to discriminate the background holes in each disocclusion and to define two spatial BG constraints that represent the lower and upper bounds of a background candidate. Holes are filled by comparing the similarities between the temporal background information and the spatial background constraints. In addition, a depth map-based ghost removal filter is described to solve the problem of the non-fit between a color image and the corresponding depth map of a virtual view. Finally, an exemplar-based inpainting is applied to fill in the remaining holes with a priority function that includes a new depth term. The experimental results demonstrated that the proposed method led to results that promised subjective and objective improvement over state-of-the-art methods.

We propose a novel coherence model for written asynchronous conversations (e.g., forums, emails), and show its applications in coherence assessment and thread reconstruction tasks.We conduct our research in two steps.First, we propose improvements to the recently proposed neural entity grid model by lexicalizing its entity transitions.Then, we extend the model to asynchronous conversations by incorporating the underlying conversational structure in the entity grid representation and feature computation.Our model achieves state of the art results on standard coherence assessment tasks in monologue and conversations outperforming existing models.We also demonstrate its effectiveness in reconstructing thread structures. *All authors contributed equally.s0: LDI Corp., Cleveland, said it will offer $50 million in commercial paper backed by leaserental receivables.s1:

We propose a local coherence model based on a convolutional neural network that operates over the entity grid representation of a text.The model captures long range entity transitions along with entity-specific features without loosing generalization, thanks to the power of distributed representation.We present a pairwise ranking method to train the model in an end-to-end fashion on a task and learn task-specific high level features.Our evaluation on three different coherence assessment tasks demonstrates that our model achieves state of the art results outperforming existing models by a good margin.

Rapid access to situation-sensitive data through social media networks creates new opportunities to address a number of real-world problems. Damage assessment during disasters is a core situational awareness task for many humanitarian organizations that traditionally takes weeks and months. In this work, we analyze images posted on social media platforms during natural disasters to determine the level of damage caused by the disasters. We employ state-of-the-art machine learning techniques to perform an extensive experimentation of damage assessment using images from four major natural disasters. We show that the domain-specific fine-tuning of deep Convolutional Neural Networks (CNN) outperforms other state-of-the-art techniques such as Bag-of-Visual-Words (BoVW). High classification accuracy under both event-specific and cross-event test settings demonstrate that the proposed approach can effectively adapt deep-CNN features to identify the severity of destruction from social media images taken after a disaster strikes.

This paper presents a hexapod crawling robot which has bioinspired design and locomotion posture from the insects. The robot, called S-Hex II, is an upgraded version of the different printable monolithic hexapod robot which is named as S-Hex I. The S-Hex II is studied for the project of developing mesoscale walking robots actuated by the soft dielectric elastomer actuators. In comparison with the S-Hex I, the S-Hex II owns smaller dimension, lighter weight, and significantly faster walking speed. We improve and increase the total number of degree-of-freedom (DOF) of the soft dielectric elastomer actuators (DEAs) from three, used in the S-Hex I, up to five, employed in the S-Hex II, and that provides the promising versatile locomotion ability to the S-Hex II robot. This robot has successfully demonstrated the back and forth ambulation on flat surfaces using the alternating tripod gait with the walking speed of 52 mm/s (approximately 0.35 body-lengths per second) at 7 Hz of actuation frequency.

This paper demonstrates a new haptic display based on soft actuator and soft sensor. The device consists of an array rigid coupling actuator and an array tactile sensor which configured sensor located on the top of actuator. The actuator includes a frame with rigid coupling made by silicone and a dielectric elastomer actuator(DEA). The movement of the DEA is transferred to touch pad via rigid coupling. Thus, it provides a soft, comport touch feeling and eliminates the danger of applying high voltage to the human skin. The actuator can work at a wide range frequency of 0–150 Hz and produce sufficient force of 50 mN over the human hand threshold. The tactile sensor can measure the pressure, locate the objects and send the feedback signals to control the actuator. In this work, a 8×12 haptic tactile display is made with high resolution. In addition, the high voltage signal processing also developed to generate variable 0–3.5 kV, and can control individual cell by using the dynamic scanning actuation algorithm. Then, it can display any shape such as a circle, a smile/sad face, a star, etc. In the future, we will build the “emotion touch pad” system which can transfer a person's emotion to another one via internet.

This paper presents a highly sensitive dual mode tactile and proximity sensor for robotic applications that uses Carbon Microcoils (CMCs). The sensor consists of multiple electrode layers printed on a Flexible Printed Circuit Board (FPCB) and a dielectric substrate into which the CMCs are dispersed. The dielectric layer is simply put on the top of the FPCB. Thus, the sensor provides ease of fabrication and robustness against repetitive external contacts because the dielectric layer protects the electrodes. The electrical properties of the sensor are changed when an object approaches or touches the sensor. The sensor uses a capacitive sensing mode for tactile sensing and an inductive sensing mode for proximity sensing. CMCs amplify the change of the sensor signal because of electrical impedance formed by the CMCs, and thus, the sensitivity of the sensor increases. We fabricate the prptotype sensor with the dimensions of 30 × 30 × 0.6 mm3, and with 3 mm spatial-resolution. The sensor detects the applied pressure up to 330 kPa and the distance of an object as much as 150 mm away.

Aiming to apply soft actuators in driving a walking robot, the design, fabrication and locomotion of a bio-inspired printable hexapod robot are studied. The robot mimics the insect's design and walking posture by driving six legs with alternating tripod gait which provides its locomotive adaptability on flat terrains. The versatile movements of the robot's leg are achieved by using soft and multiple degree-of-freedom actuators. The actuators are made by dielectric elastomers with a simple mechanism based on antagonistic configuration. By using 3D printing method, the actuator can be embedded into the frame of the robot and a control system is developed. Finally, the robot's locomotion is successfully demonstrated with variable speeds and stride lengths.

In this paper, we present a smart handheld surgical tool with tactile feedback to help microsurgery. This device consists of force sensors attached to the tool tip and tactile displays on the position of the fingerhold. The sensor is capacitive one and measures not only the magnitude of force, but also the direction of force. The sensor can measure 9N of lateral forces along x-y-z direction. The tactile display is based on dielectric elastomer actuators and made up of three rigid coupling tactile display modules. The tactile displays provide stimuli of 170μm displacement and frequency of 10Hz at the surgeons fingertips. We fabricate the device and its performances are experimentally validated.

We conducted a series of experiments to investigate effects of scale changes on distance perception in virtual environments. All experiments were carried out in an HMD. Participants first made distance estimates with feedback in a virtual tunnel (adaptation) and then made distance estimates without feedback in a differently-scaled virtual environment (test). We examined several types of scale changes, including changing the size of (1) the tunnel, (2) the targets, and (3) the separation of the two targets. Changes in target size always affected distance estimates at test. When the targets became smaller, participants overshot distance and when the targets became larger, participants undershot distance. Changes in the size of the tunnel or the separation between the targets (without a change in the size of the targets) had a minimal effect on distance estimates. These results indicate that distance estimates at test were strongly influenced by familiar size cues for distance. The discussion focuses on the stability of calibration processes and mechanisms for cue integration for perceiving distance in virtual environments.

We conducted two experiments that compared distance perception in real and virtual environments in six visual presentation methods using either timed imagined walking or direct blindfolded walking, while controlling for several other factors that could potentially impact distance perception. Our presentation conditions included unencumbered real world, real world seen through an HMD, virtual world seen through an HMD, augmented reality seen through an HMD, virtual world seen on multiple, large immersive screens, and photo-based presentation of the real world seen on multiple, large immersive screens. We found that there was a similar degree of underestimation of distance in the HMD and large-screen presentations of virtual environments. We also found that while wearing the HMD can cause some degree of distance underestimation, this effect depends on the measurement protocol used. Finally, we found that photo-based presentation did not help to improve distance perception in a large-screen immersive display system. The discussion focuses on points of similarity and difference with previous work on distance estimation in real and virtual environments.