This study examines the fairness of human- and AI-generated summaries of student reflections in university STEM classes, focusing on potential gender biases. Using topic modeling, we first identify topics that are mor...
详细信息
There is a scarcity of multilingual vision-language models that properly account for the perceptual differences that are reflected in image captions across languages and cultures. In this work, through a multimodal, m...
详细信息
Under-resourced automatic speech recognition (ASR) has become an active field of research and has experienced significant progress during the past decade. However, the performance of under-resourced ASR trained by exi...
详细信息
作者：
Hicks, AlexShi, YangLekshmi-Narayanan, Arun-BalajieeYan, WeiMarwan, SamihaDept of Computer Science
Virginia Tech Blacksburg VA Dept of Computer Science Utah State University Logan UT Intelligent Systems Program University of Pittsburgh Pittsburgh PA School of Informatics Computing and Cyber Systems North Arizona University Flagstaff AZ Dept. of Computer Science University of Virginia wCharlottesville VA
Students’ interactions while solving problems in learning environments (i.e. log data) are often used to support students’ learning. For example, researchers use log data to develop systems that can provide students...
详细信息
This paper discusses intelligent constellation generation based on autoencoder communication system. In previous studies, the amplitude was set to fluctuate between r=0.0 and 1.0. However, when checking the generated ...
详细信息
ISBN:
(纸本)9798350305142
This paper discusses intelligent constellation generation based on autoencoder communication system. In previous studies, the amplitude was set to fluctuate between r=0.0 and 1.0. However, when checking the generated constellation, distortion was confirmed instead of the conventional symbol arrangement. Therefore, in this paper, it compares the case where the amplitude is constant, the case where the average amplitude within a Minibatch is 1, and the case where the average amplitude is 1 for Interval time. The communication standard used in this research is IEEE 802.11a, assuming wireless Local Area Network (LAN) specifications. The IEEE 802.11a standard has an Fast Fourier Transform (FFT) length of 64, a subcarrier number of 52, and Quadrature Phase Shift Keying (QPSK) and 16 Quadrature Amplitude Modulation (QAM), modulation methods. A guard interval of 800 ns is added and the symbol length is 4000 ns. First, a simulation was performed under the condition that the amplitude was kept constant. QPSK with 4 symbols, constant amplitude model is rounded more than previous research result. 16QAM with 16 symbols is arranged regularly like lined up on a line. Second, the simulation was performed under the condition that the average amplitude within the minibatch was set to 1. QPSK with 4 symbols, appears to rotate clockwise. 16QAM with 16 symbols has a more uniform symbol placement than previous research result. Third, a simulation was performed under the condition that the average amplitude within Interval time was set to 1. QPSK with 4 symbols, is the closest to square among QPSK output results so far. The direction is slightly tilted, but if it can be rotated a little more, it may be possible to reproduce the same symbol arrangement as before. 16QAM with 16 symbols, the symbol arrangement is biased as a whole. However, it can be seen that are arranged in line on the line, perhaps due to regularity. As future work, in addition to the conditions set this time, it will exa
This paper proposed Scalability in Autoencoder-based Orthogonal Frequency Division Multiplexing(OFDM) communication system. In the previous research, only the comparison between IEEE802.11a and Autoencoder by the conv...
详细信息
There are many ways to describe, name, and group objects when captioning an image. Differences are evident when speakers come from diverse cultures due to the unique experiences that shape perception. Machine translat...
详细信息
Tactile sensing, which relies on direct physical contact, is critical for human perception and underpins applications in computer vision, robotics, and multimodal learning. Because tactile data is often scarce and cos...
详细信息
Existing object recognition models have been shown to lack robustness in diverse geographical scenarios due to domain shifts in design and context. Class representations need to be adapted to more accurately reflect a...
详细信息
ISBN:
(数字)9798350353006
ISBN:
(纸本)9798350353013
Existing object recognition models have been shown to lack robustness in diverse geographical scenarios due to domain shifts in design and context. Class representations need to be adapted to more accurately reflect an object concept under these shifts. In the absence of training data from target geographies, we hypothesize that geographically diverse descriptive knowledge of categories can enhance robustness. For this purpose, we explore the feasibility of probing a large language model for geography-based object knowledge, and we examine the effects of integrating knowledge into zero-shot and learnable soft prompting with CLIP. Within this exploration, we propose geog-raphy knowledge regularization to ensure that soft prompts trained on a source set of geographies generalize to an un-seen target set. Accuracy gains over prompting baselines on DollarStreet while training only on Europe data are up to +2.8/1.2/1.6 on target data from Africa/Asia/Americas, and +4.6 overall on the hardest classes. Competitive performance is shown vs. few-shot target training, and analysis is provided to direct future study of geographical robustness.
Despite recent attention to depth for various tasks, it is still an unexplored modality for weakly-supervised object detection (WSOD). We propose an amplifier method for enhancing the performance of WSOD by integratin...
Despite recent attention to depth for various tasks, it is still an unexplored modality for weakly-supervised object detection (WSOD). We propose an amplifier method for enhancing the performance of WSOD by integrating depth information. Our approach can be applied to different WSOD methods based on multiple-instance learning, without necessitating additional annotations or inducing large computational cost. Our proposed method employs monocular depth estimation to obtain hallucinated depth information, which is then incorporated into a Siamese WSOD network using contrastive loss and fusion. By analyzing the relationship between language context and depth, we calculate depth priors to identify the bounding box proposals that may contain an object of interest. These depth priors are then utilized to update the list of pseudo ground-truth boxes, or adjust the confidence of per-box predictions. We evaluate our proposed method on three datasets (COCO, PASCAL VOC, and Conceptual Captions) by implementing it on top of two state-of-the-art WSOD methods, and we demonstrate a substantial enhancement in performance.
暂无评论