Inspired by the success of autoregressive priors in probabilistic generative models,weexamineautoregressive,hierarchical, andcombined priorsasalternatives, weighing their costs and benefits in the context of image compression.

Classical GNNs excel at graph learning but incur high computational costs in large-scale settings. We present a fully quantum Graph Neural Network (QGNN) that implements message passing via Parameterized Quantum Circuits (PQCs). Our Quantum Graph Convolutional Layers (QGCLs) encode features into quantum states, process graphs with NISQ-compatible unitaries, and retrieve embeddings through measurement. Applied to D2D power control for SINR maximization, our QGNN matches classical performance with fewer parameters and inherent parallelism. This end-to-end PQC-based GNN marks a step toward quantum-accelerated wireless optimization.

Determining the minimum width of fully connected neural networks has become a fundamental problem in recent theoretical studies of deep neural networks. In this paper, we study the lower bounds and upper bounds of the minimum width required for fully connected neural networks in order to have universal approximation capability, which is important in network design and training. We show that $w_{min}\leq\max(2d_x+1, d_y)$ also holds true for networks with ELU, SELU activation functions, and the upper bound of this inequality is attained when $d_y=2d_x$, where $d_x$, $d_y$ denote the input and output dimensions, respectively. Besides, we show that $d_x+1\leq w_{min}\leq d_x+d_y$ for networks with LeakyReLU, ELU, CELU, SELU, Softplus activation functions, by proving that ReLU activation function can be approximated by these activation functions. In addition, in the case that the activation function is injective or can be uniformly approximated by a sequence of injective functions (e.g., ReLU), we present a new proof of the inequality $w_{min}\ge d_y+\mathbf{1}_{d_x

We consider the contextual combinatorial bandit setting where in each round, the learning agent, e.g., a recommender system, selects a subset of "arms," e.g., products, and observes rewards for both the individual base arms, which are a function of known features (called "context"), and the super arm (the subset of arms), which is a function of the base arm rewards. The agent's goal is to simultaneously learn the unknown reward functions and choose the highest-reward arms. For example, the "reward" may represent a user's probability of clicking on one of the recommended products. Conventional bandit models, however, employ restrictive reward function models in order to obtain performance guarantees. We make use of deep neural networks to estimate and learn the unknown reward functions and propose Neural UCB Clustering (NeUClust), which adopts a clustering approach to select the super arm in every round by exploiting underlying structure in the context space. Unlike prior neural bandit works, NeUClust uses a neural network to estimate the super arm reward and select the super arm, thus eliminating the need for a known optimization oracle. We non-trivially extend prior neural combinatorial bandit works to prove that NeUClust achieves $\widetilde{O}\left(\widetilde{d}\sqrt{T}\right)$ regret, where $\widetilde{d}$ is the effective dimension of a neural tangent kernel matrix, $T$ the number of rounds. Experiments on real world recommendation datasets show that NeUClust achieves better regret and reward than other contextual combinatorial and neural bandit algorithms.

Traditional methods for human localization and pose estimation (HPE), which mainly rely on RGB images as an input modality, confront substantial limitations in real-world applications due to privacy concerns. In contrast, radar-based HPE methods emerge as a promising alternative, characterized by distinctive attributes such as through-wall recognition and privacy-preserving, rendering the method more conducive to practical deployments. This paper presents a Radar Tensor-based human pose (RT-Pose) dataset and an open-source benchmarking framework. The RT-Pose dataset comprises 4D radar tensors, LiDAR point clouds, and RGB images, and is collected for a total of 72k frames across 240 sequences with six different complexity-level actions. The 4D radar tensor provides raw spatio-temporal information, differentiating it from other radar point cloud-based datasets. We develop an annotation process using RGB images and LiDAR point clouds to accurately label 3D human skeletons. In addition, we propose HRRadarPose, the first single-stage architecture that extracts the high-resolution representation of 4D radar tensors in 3D space to aid human keypoint estimation. HRRadarPose outperforms previous radar-based HPE work on the RT-Pose benchmark. The overall HRRadarPose performance on the RT-Pose dataset, as reflected in a mean per joint position error (MPJPE) of 9.91cm, indicates the persistent challenges in achieving accurate HPE in complex real-world scenarios. RT-Pose is available at https://huggingface.co/datasets/uwipl/RT-Pose.

This study proposes a framework for enhancing the stroke quality of badminton players by generating personalized motion guides, utilizing a multimodal wearable dataset. These guides are based on counterfactual algorithms and aim to reduce the performance gap between novice and expert players. Our approach provides joint-level guidance through visualizable data to assist players in improving their movements without requiring expert knowledge. The method was evaluated against a traditional algorithm using metrics to assess validity, proximity, and plausibility, including arithmetic measures and motion-specific evaluation metrics. Our evaluation demonstrates that the proposed framework can generate motions that maintain the essence of original movements while enhancing stroke quality, providing closer guidance than direct expert motion replication. The results highlight the potential of our approach for creating personalized sports motion guides by generating counterfactual motion guidance for arbitrary input motion samples of badminton strokes.

A recurrent neural network (RNN) is a widely used deep-learning network for dealing with sequential data. Imitating a dynamical system, an infinite-width RNN can approximate any open dynamical system in a compact domain. In general, deep networks with bounded widths are more effective than wide networks in practice; however, the universal approximation theorem for deep narrow structures has yet to be extensively studied. In this study, we prove the universality of deep narrow RNNs and show that the upper bound of the minimum width for universality can be independent of the length of the data. Specifically, we show that a deep RNN with ReLU activation can approximate any continuous function or $L^p$ function with the widths $d_x+d_y+2$ and $\max\{d_x+1,d_y\}$, respectively, where the target function maps a finite sequence of vectors in $\mathbb{R}^{d_x}$ to a finite sequence of vectors in $\mathbb{R}^{d_y}$. We also compute the additional width required if the activation function is $\tanh$ or more. In addition, we prove the universality of other recurrent networks, such as bidirectional RNNs. Bridging a multi-layer perceptron and an RNN, our theory and proof technique can be an initial step toward further research on deep RNNs.

We demonstrate the conversational Chatbot platform named Chatti which supports developers with a tool to develop their chatbot easily without full understanding technologies inside a conversational chatbot. To develop a chatbot with Chatti, a developer inputs customized domain data and deploys his Chatbot with a tool. Then users can interact with the Chatbot based on natural language conversation via messengers and so on. Chatti includes natural language understanding, dialog management, action planning, natural language generation and chitchat component which run on g models learned from developers' input data as in common in conversational assistants such as Bixby, Siri, Alexa and etc. With Chatti, the developer could make his Chatbot support two types of conversation simultaneously – basic chitchat and task-oriented dialog. In contrast to prior chatbot building tools are mainly focused on the Natural Language Understanding, Chatti is more focused on full dialog system – dialog management, action planning, natural language generation and chitchat. We believe Chatti could accelerate a wide possibility of conversational Chatbot for services as well as IoT devices.

In retrospect, Nisos experts made the right forecast. However, this was a clear minority opinion. Before and after their report, dozens of politicians and institutions drew considerable attention to the approaching danger: 'imagine a scenario where, on the eve of next year's presidential election, the Democratic nominee appears in a video where he or she endorses President Trump. Now, imagine it the other way around.' (Sprangler, 2019). It is fair to say that deepfakes' high potential for disinformation was noticed long before these hypothetical consequences were evoked, mainly because they were revealed to be highly credible. Two examples: 'In an online quiz, 49 percent of people who visited our site said they incorrectly believed Nixon's synthetically altered face was real and 65 percent thought his voice was real' (Panetta et al, 2020), or'Two-thirds of participants believed that one day it would be impossible to discern a real video from a fake one.

Researchers discussed how they used a deep-learning algorithm to detect, quantify, and assess the severity of infarcts in the brain on diffusion-weighted MRI (DWI-MRI) exams in acute ischemic stroke patients in a Sunday presentation at the virtual RSNA 2020 meeting. A team of researchers led by presenter Seung Hyun Hwang of Yonsei University in Seoul, South Korea, developed a deep-learning model that can segment and quantify brain infarcts using DWI-MRI and then assess their severity by analyzing apparent diffusion coefficient (ADC) maps of the lesions. In testing, the model achieved high sensitivity and specificity. "The qualitative and quantitative results of our study show feasibility for detecting and quantifying infarcts," Hwang said. Due to its sensitivity for the detection of small and early infarcts, DWI-MRI is commonly used for evaluation of acute ischemic stroke, according to Hwang.