arXiv (Cornell University)最新文献_第2页

Weak expansion properties and a large deviation principle for coarse expanding conformal systems 粗膨胀保形系统的弱膨胀特性和大偏差原理

arXiv (Cornell University) Pub Date : 2023-11-13 DOI: 10.48550/arxiv.2311.07305

Li, Zhiqiang, Zheng, Hanyun

引用次数: 0

HESS J1809-193: Gamma-Ray Emission by Cosmic Rays from Past Explosion HESS J1809-193:由过去爆炸的宇宙射线发射的伽马射线

arXiv (Cornell University) Pub Date : 2023-11-13 DOI: 10.48550/arxiv.2311.07481

Boxi, Sovan, Gupta, Nayantara

引用次数: 0

Temporal Performance Prediction for Deep Convolutional Long Short-Term Memory Networks 深度卷积长短期记忆网络的时间性能预测

arXiv (Cornell University) Pub Date : 2023-11-13 DOI: 10.48550/arxiv.2311.07477

Fieback, Laura, Dash, Bidya, Spiegelberg, Jakob, Gottschalk, Hanno

引用次数: 0

Are We Falling in a Middle-Intelligence Trap? An Analysis and Mitigation of the Reversal Curse 我们是否陷入了中等智力的陷阱?逆转咒的分析与缓解

arXiv (Cornell University) Pub Date : 2023-11-13 DOI: 10.48550/arxiv.2311.07468

Lv, Ang, Zhang, Kaiyi, Xie, Shufang, Tu, Quan, Chen, Yuhan, Wen, Ji-Rong, Yan, Rui

{"title":"Are We Falling in a Middle-Intelligence Trap? An Analysis and Mitigation\u0000 of the Reversal Curse","authors":"Lv, Ang, Zhang, Kaiyi, Xie, Shufang, Tu, Quan, Chen, Yuhan, Wen, Ji-Rong, Yan, Rui","doi":"10.48550/arxiv.2311.07468","DOIUrl":"https://doi.org/10.48550/arxiv.2311.07468","url":null,"abstract":"Recent studies have highlighted a phenomenon in large language models (LLMs) known as \"the reversal curse,\" in which the order of knowledge entities in the training data biases the models' comprehension. For example, if a model is trained on sentences where entity A consistently appears before entity B, it can respond to queries about A by providing B. However, it may encounter confusion when presented with questions concerning B. We contend that the reversal curse is partially a result of specific model training objectives, particularly evident in the prevalent use of the next-token prediction within most causal language models. For the next-token prediction, models solely focus on a token's preceding context, resulting in a restricted comprehension of the input. In contrast, we illustrate that the GLM, trained using the autoregressive blank infilling objective where tokens to be predicted have access to the entire context, exhibits better resilience against the reversal curse. We propose a novel training method, BIdirectional Casual language modeling Optimization (BICO), designed to mitigate the reversal curse when fine-tuning pretrained causal language models on new data. BICO modifies the causal attention mechanism to function bidirectionally and employs a mask denoising optimization. In the task designed to assess the reversal curse, our approach improves Llama's accuracy from the original 0% to around 70%. We hope that more attention can be focused on exploring and addressing these inherent weaknesses of the current LLMs, in order to achieve a higher level of intelligence.","PeriodicalId":496270,"journal":{"name":"arXiv (Cornell University)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"136352506","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 1

Finding and Editing Multi-Modal Neurons in Pre-Trained Transformer 在预训练变压器中寻找和编辑多模态神经元

arXiv (Cornell University) Pub Date : 2023-11-13 DOI: 10.48550/arxiv.2311.07470

Pan, Haowen, Cao, Yixin, Wang, Xiaozhi, Yang, Xun

引用次数: 0

Goldberger-Treiman relation and Wu-type experiment in the decuplet sector 双扇区的Goldberger-Treiman关系和wu型实验

arXiv (Cornell University) Pub Date : 2023-11-13 DOI: 10.48550/arxiv.2311.07471

Bertilsson, Magnus, Leupold, Stefan

{"title":"Goldberger-Treiman relation and Wu-type experiment in the decuplet\u0000 sector","authors":"Bertilsson, Magnus, Leupold, Stefan","doi":"10.48550/arxiv.2311.07471","DOIUrl":"https://doi.org/10.48550/arxiv.2311.07471","url":null,"abstract":"The leading-order chiral Lagrangian for the baryon octet and decuplet states coupled to Goldstone bosons and external sources contains six low-energy constants. Five of them are fairly well known from phenomenology, but the sixth one is practically unknown. This coupling constant provides the strength of the (p-wave) coupling of Goldstone bosons to decuplet states. Its size and even sign are under debate. Quark model and QCD for a large number of colors provide predictions, but some recent phenomenological analyses suggest even an opposite sign for the Delta-pion coupling. The Goldberger-Treiman relation connects this coupling constant to the axial charge of the Delta baryon. This suggests a Wu-type experiment to determine the unknown low-energy constant. While this is not feasible in the Delta sector because of the large hadronic width of the Delta, there is a flavor symmetry related process that is accessible: the weak semileptonic decay of the Omega baryon to a spin 3/2 cascade baryon. A broad research program is suggested that can pin down at least the rough size and the sign of the last unknown low-energy constant of the leading-order Lagrangian. It encompasses experimental measurements, in particular the forward-backward asymmetry of the semileptonic decay, together with a determination of the quark-mass dependences using lattice QCD for the narrow decuplet states and chiral perturbation theory to extrapolate to the Delta sector. Besides discussing the strategy of the research program, the present work provides a feasibility check based on a simple leading-order calculation.","PeriodicalId":496270,"journal":{"name":"arXiv (Cornell University)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"136352511","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

The apparent tidal decay of WASP-4 b can be explained by the R{o}mer effect wasp - 4b的明显潮汐衰减可以用R{o}mer效应来解释

arXiv (Cornell University) Pub Date : 2023-11-13 DOI: 10.48550/arxiv.2311.07098

Harre, Jan-Vincent, Smith, Alexis M. S.

{"title":"The apparent tidal decay of WASP-4 b can be explained by the R{o}mer\u0000 effect","authors":"Harre, Jan-Vincent, Smith, Alexis M. S.","doi":"10.48550/arxiv.2311.07098","DOIUrl":"https://doi.org/10.48550/arxiv.2311.07098","url":null,"abstract":"Tidal orbital decay plays a vital role in the evolution of hot Jupiter systems. As of now, this was only observationally confirmed for the WASP-12 system. There are a few other candidates, including WASP-4 b, but no conclusive result could be obtained for these systems as of yet. In this study, we present an analysis of new TESS data of WASP-4 b together with archival data, taking the light-time effect (LTE), induced by the second planetary companion, into account as well. We make use of three different Markov-Chain-Monte-Carlo models; a circular orbit with a constant orbital period, a circular orbit with a decaying orbit, and an elliptical orbit with apsidal precession. This analysis is repeated for four cases. The first case features no LTE correction, with the remaining three cases featuring three different timing correction approaches. Comparison of these models yields no conclusive answer to the cause of WASP-4,b's apparent transit timing variations. A broad range of values of the orbital decay and apsidal precession parameters are possible, depending on the LTE correction. This work highlights the importance of continued photometric and spectroscopic monitoring of hot Jupiters.","PeriodicalId":496270,"journal":{"name":"arXiv (Cornell University)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"136352634","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Collaborative Goal Tracking of Multiple Mobile Robots Based on Geometric Graph Neural Network 基于几何图神经网络的多移动机器人协同目标跟踪

arXiv (Cornell University) Pub Date : 2023-11-13 DOI: 10.48550/arxiv.2311.07105

Lin, Qingquan, Lu, Weining

{"title":"Collaborative Goal Tracking of Multiple Mobile Robots Based on Geometric\u0000 Graph Neural Network","authors":"Lin, Qingquan, Lu, Weining","doi":"10.48550/arxiv.2311.07105","DOIUrl":"https://doi.org/10.48550/arxiv.2311.07105","url":null,"abstract":"Multi-robot systems are widely used in spatially distributed tasks, and their collaborative path planning is of great significance for working efficiency. Currently, different multi-robot collaborative path planning methods have been proposed, but how to process the sensory information of neighboring robots at different locations from a local perception perspective in real environment to make better decisions is still a major difficulty. To address this problem, this paper proposes a multi-robot collaborative path planning method based on geometric graph neural network (GeoGNN). GeoGNN introduces the relative position information of neighboring robots into each interaction layer of the graph neural network to better integrate neighbor sensing information. An expert data generation method is designed for the robot to advance in a single step, by which expert data are generated in ROS to train the network. Experimental results show that the accuracy of the proposed method is improved by about 5% compared to the model based only on CNN on the expert data set. In ROS simulation environment path planning test, the success rate is improved by about 4% compared to CNN and flowtime increase is reduced about 8%, which outperforms other graph neural network models.","PeriodicalId":496270,"journal":{"name":"arXiv (Cornell University)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"136352641","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Secure Wireless Communication via Movable-Antenna Array 移动天线阵列的安全无线通信

arXiv (Cornell University) Pub Date : 2023-11-13 DOI: 10.48550/arxiv.2311.07104

Hu, Guojie, Wu, Qingqing, Xu, Kui, Si, Jiangbo, Al-Dhahir, Naofal

引用次数: 0

Robust Probabilistic Prediction for Stochastic Dynamical Systems 随机动力系统的鲁棒概率预测

arXiv (Cornell University) Pub Date : 2023-11-13 DOI: 10.48550/arxiv.2311.07108

Xu, Tao, He, Jianping

{"title":"Robust Probabilistic Prediction for Stochastic Dynamical Systems","authors":"Xu, Tao, He, Jianping","doi":"10.48550/arxiv.2311.07108","DOIUrl":"https://doi.org/10.48550/arxiv.2311.07108","url":null,"abstract":"It is critical and challenging to design robust predictors for stochastic dynamical systems (SDSs) with uncertainty quantification (UQ) in the prediction. Specifically, robustness guarantees the worst-case performance when the predictor's information set of the system is inadequate, and UQ characterizes how confident the predictor is about the predictions. However, it is difficult for traditional robust predictors to provide robust UQ because they were designed to robustify the performance of point predictions. In this paper, we investigate how to robustify the probabilistic prediction for SDS, which can inherently provide robust distributional UQ. To characterize the performance of probabilistic predictors, we generalize the concept of likelihood function to likelihood functional, and prove that this metric is a proper scoring rule. Based on this metric, we propose a framework to quantify when the predictor is robust and analyze how the information set affects the robustness. Our framework makes it possible to design robust probabilistic predictors by solving functional optimization problems concerning different information sets. In particular, we design a class of moment-based optimal robust probabilistic predictors and provide a practical Kalman-filter-based algorithm for implementation. Extensive numerical simulations are provided to elaborate on our results.","PeriodicalId":496270,"journal":{"name":"arXiv (Cornell University)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"136352643","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0