检索结果-内蒙古大学图书馆

GrounDial: Human-norm Grounded Safe Dialog Response Generation

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Kim, Siwon Dai, Shuyang Kachuee, Mohammad Ray, Shayan Taghavi, Tara Yoon, Sungroh Department of ECE Seoul National University Korea Republic of Amazon United States Interdisciplinary Program in AI Seoul National University Korea Republic of

Current conversational ai systems based on large language models (LLMs) are known to generate unsafe responses, agreeing to offensive user input or including toxic content. Previous research aimed to alleviate the toxicity, by fine-tuning LLM with manually annotated safe dialogue histories. However, the dependency on additional tuning requires substantial costs. To remove the dependency, we propose GrounDial, where response safety is achieved by grounding responses to commonsense social rules without requiring fine-tuning. A hybrid approach of in-context learning and human-norm-guided decoding of GrounDial enables responses to be quantitatively and qualitatively safer even without additional data or tuning. © 2024, CC BY-NC-SA.

关键词：

SemanticDraw: Towards Real-Time Interactive Content Creation from Image Diffusion Models

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Lee, Jaerin Jung, Daniel Sungho Lee, Kanggeon Lee, Kyoung Mu ASRI Department of ECE Interdisciplinary Program in Artificial Intelligence Korea Republic of SNU-LG AI Research Center Seoul National University Korea Republic of

We introduce SemanticDraw, a new paradigm of interactive content creation where high-quality images are generated in near real-time from given multiple hand-drawn regions, each encoding prescribed semantic meaning. In order to maximize the productivity of content creators and to fully realize their artistic imagination, it requires both quick interactive interfaces and fine-grained regional controls in their tools. Despite astonishing generation quality from recent diffusion models, we find that existing approaches for regional controllability are very slow (52 seconds for 512 × 512 image) while not compatible with acceleration methods such as LCM, blocking their huge potential in interactive content creation. From this observation, we build our solution for interactive content creation in two steps: (1) we establish compatibility between region-based controls and acceleration techniques for diffusion models, maintaining high fidelity of multi-prompt image generation with×10 reduced number of inference steps, (2) we increase the generation throughput with our new multi-prompt stream batch pipeline, enabling low-latency generation from multiple, region-based text prompts on a single RTX 2080 Ti GPU. Our proposed framework is generalizable to any existing diffusion models and acceleration schedulers, allowing sub-second (0.64 seconds) image content creation application upon well-established image diffusion models. Our project page is: https: //***/research/semantic-draw. Copyright © 2024, The Authors. All rights reserved.

关键词： Semantics

Paralinguistics-aware speech-empowered large language models for natural conversation 24

学校读者我要写书评

暂无评论

Paralinguistics-aware speech-empowered large language models...

Proceedings of the 38th International Conference on Neural Information Processing Systems

作者： Heeseung Kim Soonshin Seo Kyeongseok Jeong Ohsung Kwon Soyoon Kim Jungwhan Kim Jaehong Lee Eunwoo Song Myungwoo Oh Jung-Woo Ha Sungroh Yoon Kang Min Yoo Data Science and AI Lab Department of ECE Seoul National University NAVER Cloud NAVER Cloud and Artificial Intelligence Institute Seoul National University NAVER Cloud and NAVER AI Lab Data Science and AI Lab Department of ECE Seoul National University and Artificial Intelligence Institute Seoul National University and ASRI INMC ISRC and Interdisciplinary Program in AI Seoul National University NAVER Cloud and NAVER AI Lab and Artificial Intelligence Institute Seoul National University

ISBN: (纸本)9798331314385

关键词：

Paralinguistics-Aware Speech-Empowered Large Language Models for Natural Conversation

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Kim, Heeseung Seo, Soonshin Jeong, Kyeongseok Kwon, Ohsung Kim, Soyoon Kim, Jungwhan Lee, Jaehong Song, Eunwoo Oh, Myungwoo Ha, Jung-Woo Yoon, Sungroh Yoo, Kang Min Data Science and AI Lab Department of ECE Seoul National University Korea Republic of NAVER Cloud NAVER AI Lab Artificial Intelligence Institute Seoul National University Korea Republic of ASRI INMC ISRC Interdisciplinary Program in AI Seoul National University Korea Republic of

关键词： Semantics

Notice of Retraction: E2V-SDE: From Asynchronous Events to Fast and Continuous Video Reconstruction via Neural Stochastic Differential Equations

学校读者我要写书评

暂无评论

Notice of Retraction: E2V-SDE: From Asynchronous Events to F...

Conference on Computer Vision and Pattern Recognition (CVPR)

作者： Jongwan Kim DongJin Lee Byunggook Na Seongsik Park Jeonghee Jo Sungroh Yoon Interdisciplinary Program in AI Seoul National University Department of ECE Seoul National University Korea Institute of Science and Technology AIIS ASRI INMC ISRC Seoul National University

Retracted.

ISBN: (数字)9781665469463

ISBN: (纸本)9781665469470

Retracted.

关键词：

Demystifying the Neural Tangent Kernel from a Practical Perspective: Can it be trusted for Neural Architecture Search without training?

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Mok, Jisoo Na, Byunggook Kim, Ji-Hoon Han, Dongyoon Yoon, Sungroh Department of ECE Seoul National University Korea Republic of NAVER AI Lab Korea Republic of NAVER CLOVA Korea Republic of AIIS ASRI INMC ISRC Interdisciplinary Program in AI Seoul National University Korea Republic of

In Neural Architecture Search (NAS), reducing the cost of architecture evaluation remains one of the most crucial challenges. Among a plethora of efforts to bypass training of each candidate architecture to convergence for evaluation, the Neural Tangent Kernel (NTK) is emerging as a promising theoretical framework that can be utilized to estimate the performance of a neural architecture at initialization. In this work, we revisit several at-initialization metrics that can be derived from the NTK and reveal their key shortcomings. Then, through the empirical analysis of the time evolution of NTK, we deduce that modern neural architectures exhibit highly non-linear characteristics, making the NTK-based metrics incapable of reliably estimating the performance of an architecture without some amount of training. To take such non-linear characteristics into account, we introduce Label-Gradient Alignment (LGA), a novel NTK-based metric whose inherent formulation allows it to capture the large amount of non-linear advantage present in modern neural architectures. With minimal amount of training, LGA obtains a meaningful level of rank correlation with the final test accuracy of an architecture. Lastly, we demonstrate that LGA, complemented with few epochs of training, successfully guides existing search algorithms to achieve competitive search performances with significantly less search cost. The code is available at: https://***/nutellamok/DemystifyingNTK. © 2022, CC BY-NC-ND.

关键词：

Confidence Score for Source-Free Unsupervised Domain Adaptation

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Lee, Jonghyun Jung, Dahuin Yim, Junho Yoon, Sungroh Data Science and AI Lab. Seoul National University Korea Republic of AIRS Company Hyundai Motor Group Seoul Korea Republic of Department of ECE Interdisciplinary Program in AI National University Seoul Korea Republic of

Source-free unsupervised domain adaptation (SFUDA) aims to obtain high performance in the unlabeled target domain using the pre-trained source model, not the source data. Existing SFUDA methods assign the same importance to all target samples, which is vulnerable to incorrect pseudo-labels. To differentiate between sample importance, in this study, we propose a novel sample-wise confidence score, the Joint Model-Data Structure (JMDS) score for SFUDA. Unlike existing confidence scores that use only one of the source or target domain knowledge, the JMDS score uses both knowledge. We then propose a Confidence score Weighting Adaptation using the JMDS (CoWA-JMDS) framework for SFUDA. CoWA-JMDS consists of the JMDS scores as sample weights and weight Mixup that is our proposed variant of Mixup. Weight Mixup promotes the model make more use of the target domain knowledge. The experimental results show that the JMDS score outperforms the existing confidence scores. Moreover, CoWA-JMDS achieves state-of-the-art performance on various SFUDA scenarios: closed, open, and partial-set scenarios. Copyright © 2022, The Authors. All rights reserved.

关键词： Domain Knowledge

Guided-TTS: A Diffusion Model for Text-to-Speech via Classifier Guidance

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Kim, Heeseung Kim, Sungwon Yoon, Sungroh Data Science and AI Lab. Seoul National University Korea Republic of Department of ECE and Interdisciplinary Program in AI Seoul National University Korea Republic of

We propose Guided-TTS, a high-quality text-to-speech (TTS) model that does not require any transcript of target speaker using classifier guidance. Guided-TTS combines an unconditional diffusion probabilistic model with a separately trained phoneme classifier for classifier guidance. Our unconditional diffusion model learns to generate speech without any context from untranscribed speech data. For TTS synthesis, we guide the generative process of the diffusion model with a phoneme classifier trained on a large-scale speech recognition dataset. We present a norm-based scaling method that reduces the pronunciation errors of classifier guidance in Guided-TTS. We show that Guided-TTS achieves a performance comparable to that of the state-of-the-art TTS model, Grad-TTS, without any transcript for LJSpeech. We further demonstrate that Guided-TTS performs well on diverse datasets including a long-form untranscribed dataset. Copyright © 2021, The Authors. All rights reserved.

关键词： Large dataset

E2V-SDE: From Asynchronous Events to Fast and Continuous Video Reconstruction via Neural Stochastic Differential Equations

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Kim, Jongwan Lee, DongJin Na, Byunggook Park, Seongsik Jo, Jeonghee Yoon, Sungroh Interdisciplinary Program in AI Seoul National University Korea Republic of Department of ECE Seoul National University Korea Republic of AIIS ASRI INMC ISRC Seoul National University Korea Republic of Korea Institute of Science and Technology Korea Republic of

Event cameras respond to brightness changes in the scene asynchronously and independently for every pixel. Due to the properties, these cameras have distinct features: high dynamic range (HDR), high temporal resolution, and low power consumption. However, the results of event cameras should be processed into an alternative representation for computer vision tasks. Also, they are usually noisy and causes poor performance in areas with few events. In recent years, numerous researchers have attempted to reconstruct videos from events. However, they do not provide good quality videos due to a lack of temporal information from irregular and discontinuous data. To overcome these difficulties, we introduce an E2V-SDE whose dynamics are governed in a latent space by Stochastic differential equations (SDE). Therefore, E2V-SDE can rapidly reconstruct images at arbitrary time steps and make realistic predictions on unseen data. In addition, we successfully adopted a variety of image composition techniques for improving image clarity and temporal consistency. By conducting extensive experiments on simulated and real-scene datasets, we verify that our model outperforms state-of-the-art approaches under various video reconstruction settings. In terms of image quality, the LPIPS score improves by up to 12% and the reconstruction speed is 87% higher than that of ET-Net. Copyright © 2022, The Authors. All rights reserved.

关键词： Cameras