检索结果-内蒙古大学图书馆

Simplifying Multimodality: Unimodal Approach to Multimodal Challenges in Radiology with General-Domain Large Language Model

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Cho, Seonhee Kim, Choonghan Lee, Jiho Chilkunda, Chetan Choi, Sujin Yoon, Joo Heung Kim Jaechul Graduate School of AI KAIST Gyeonggi Seongnam Korea Republic of Moon Soul Graduate School of Future Strategy KAIST Daejeon Korea Republic of Department of Computer Science and Engineering Ewha Womans University Seoul Korea Republic of Department of Chemical Engineering Carnegie Mellon University PittsburghPA United States Division of Environmental Science and Engineering POSTECH Gyeongbuk Pohang Korea Republic of Division of Pulmonary Allergy Critical Care and Sleep Medicine Department of Medicine University of Pittsburgh PittsburghPA United States

Recent advancements in Large Multimodal Models (LMMs) have attracted interest in their generalization capability with only a few samples in the prompt. This progress is particularly relevant to the medical domain, where the quality and sensitivity of data pose unique challenges for model training and application. However, the dependency on high-quality data for effective in-context learning raises questions about the feasibility of these models when encountering with the inevitable variations and errors inherent in real-world medical data. In this paper, we introduce MID-M, a novel framework that leverages the in-context learning capabilities of a general-domain Large Language Model (LLM) to process multimodal data via image descriptions. MID-M achieves a comparable or superior performance to task-specific fine-tuned LMMs and other general-domain ones, without the extensive domain-specific training or pre-training on multimodal data, with significantly fewer parameters. This highlights the potential of leveraging general-domain LLMs for domain-specific tasks and offers a sustainable and cost-effective alternative to traditional LMM developments. Moreover, the robustness of MID-M against data quality issues demonstrates its practical utility in real-world medical domain applications. © 2024, CC BY.

关键词： Cost effectiveness

Higher-Order Color Voronoi Diagrams and the Colorful Clarkson–Shor Framework

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Bae, Sang Won Oliver, Nicolau Papadopoulou, Evanthia Division of AI Computer Science and Engineering Kyonggi University Suwon Korea Republic of Faculty of Informatics Università della Svizzera italiana Lugano Switzerland

Given a set S of n colored sites, each s ∈ S associated with a distance-to-site function δs : R2 → R, we consider two distance-to-color functions for each color: one takes the minimum of δs for sites s ∈ S in that color and the other takes the maximum. These two sets of distance functions induce two families of higher-order Voronoi diagrams for colors in the plane, namely, the minimal and maximal order-k color Voronoi diagrams, which include various well-studied Voronoi diagrams as special cases. In this paper, we derive an exact upper bound 4k(n − k) − 2n on the total number of vertices in both the minimal and maximal order-k color diagrams for a wide class of distance functions δs that satisfy certain conditions, including the case of point sites S under convex distance functions and the Lp metric for any 1 ⩽ p ⩽ ∞. For the L1 (or, L∞) metric, and other convex polygonal metrics, we show that the order-k minimal diagram of point sites has O(min{k(n− k), (n− k)2 }) complexity, while its maximal counterpart has O(min{k(n − k), k2 }) complexity. To obtain these combinatorial results, we extend the Clarkson–Shor framework to colored objects, and demonstrate its application to several fundamental geometric structures, including higher-order color Voronoi diagrams, colored j-facets, and levels in the arrangements of piecewise linear/algebraic curves/surfaces. We also present an iterative approach to compute higher-order color Voronoi diagrams. Copyright © 2025, The Authors. All rights reserved.

关键词： Graph theory

Dual-Stage Deeply Supervised Attention-based Convolutional Neural Networks for Mandibular Canal Segmentation in CBCT Scans

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Usman, Muhammad Rehman, Azka Saleem, Amal Jawaid, Rabeea Byon, Shi-Sub Kim, Sung-Hyun Lee, Byoung Dai Lee, Byung-Il Shin, Yeong-Gil Center for Artificial Intelligence in Medicine and Imaging HealthHub Co. Ltd. Seoul06524 Korea Republic of Seoul National University Seoul Korea Republic of Division of AI and Computer Engineering Kyonggi University Suwon Korea Republic of

Accurate segmentation of mandibular canals in lower jaws is important in dental implantology. Medical experts determine the implant position and dimensions manually from 3D CT images to avoid damaging the mandibular nerve inside the canal. In this paper, we propose a novel dual-stage deep learning-based scheme for the automatic segmentation of the mandibular canal. Particularly, we first enhance the CBCT scans by employing the novel histogram-based dynamic windowing scheme, which improves the visibility of mandibular canals. After enhancement, we design 3D deeply supervised attention U-Net architecture for localizing the volumes of interest (VOIs), which contain the mandibular canals (i.e., left and right canals). Finally, we employed the multi-scale input residual U-Net architecture (MS-R-UNet) to segment the mandibular canals using VOIs accurately. The proposed method has been rigorously evaluated on 500 scans. The results demonstrate that our technique outperforms the current state-of-the-art segmentation performance and robustness methods. © 2022, CC0.

关键词： Canals

The SVD of Convolutional Weights: A CNN Interpretability Framework

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Praggastis, Brenda Brown, Davis Marrero, Carlos Ortiz Purvine, Emilie Shapiro, Madelyn Wang, Bei AI & Data Analytics Division Pacific Northwest National Laboratory SeattleWA98109 United States AI & Data Analytics Division Pacific Northwest National Laboratory RichlandWA99354 United States Department of Electrical & Computer Engineering North Carolina State University RaleighNC27607 United States Institute University of Utah Salt Lake CityUT94112 United States

Deep neural networks used for image classification often use convolutional filters to extract distinguishing features before passing them to a linear classifier. Most interpretability literature focuses on providing semantic meaning to convolutional filters to explain a model’s reasoning process and confirm its use of relevant information from the input domain. Fully connected layers can be studied by decomposing their weight matrices using a singular value decomposition, in effect studying the correlations between the rows in each matrix to discover the dynamics of the map. In this work we define a singular value decomposition for the weight tensor of a convolutional layer, which provides an analogous understanding of the correlations between filters, exposing the dynamics of the convolutional map. We validate our definition using recent results in random matrix theory. By applying the decomposition across the linear layers of an image classification network we suggest a framework against which interpretability methods might be applied using hypergraphs to model class separation. Rather than looking to the activations to explain the network, we use the singular vectors with the greatest corresponding singular values for each linear layer to identify those features most important to the network. We illustrate our approach with examples and introduce the DeepDataProfiler library, the analysis tool used for this *** Codes 68T07, 68T01, 05C65 © 2022, CC BY.

关键词： Singular value decomposition

A Hybrid Approach Towards Data optimization for Cloud Based IoT Environment 4

学校读者我要写书评

暂无评论

A Hybrid Approach Towards Data optimization for Cloud Based ...

4th International Conference on Emerging Trends in Networks and computer Communications, ETNCC 2024

作者： Pal, Souvik Jhanjhi, N.Z. Akila, D. Abdullah, Monir Humayun, Memona Kumar, Sandeep Kumar, Rahul Saveetha Institute of Medical & Tech. Sciences Dept. of Management IS Saveetha College of Liberal Arts & Sciences Chennai602105 India Sister Nivedita University Dept. of CS and Engineering Kolkata700156 India School of Computer Science SCS Taylor's University Subang Jaya47500 Malaysia SIMATS Deemed University Saveetha College of Liberal Arts & Sciences Dept. of Computer Applications Chennai602105 India University of Bisha Department of CS and AI College of Computing & IT Bisha67714 Saudi Arabia School of Arts Humanities & Social Sciences University of Roehampton Department of Computing LondonSW15 5PJ United Kingdom India KKR and KSR Institute of Technology and Sciences Division of Research and Development Lovely Professional University Phagwara Andhra Pradesh Guntur144411 India

ISBN: (纸本)9798350353266

An innovative method for data optimization in the Internet of Things (IoT) assister sensing devices is presented in this research. The many elements of the cloud platform with IoT assistance are addressed. A fresh design for IoT-aided sensor networks is introduced to monitor the environment. A Data Optimisation in Cloud Internet of Things (DO-CIoT) is proposed in this paper to optimize the data stored in the cloud and presents methods to access the data in end devices. The suggested technique uses a membership function to improve data allocation while considering all node data. The suggested model is assessed regarding performance ratios, reliability percentage, coverage ratios, and sensing error and contrasted with two-stage efficiency enhancement. A mean performance ratio of 89%, reliability of 91%, coverage of 89 © 2024 IEEE.

关键词： Cloud platforms

Biomolecular and quantum algorithms for the dominating set problem in arbitrary networks

学校读者我要写书评

暂无评论

Scientific Reports 2023年第1期13卷 1-11页

作者： Wong, Renata Chang, Weng-Long Chung, Wen-Yu Vasilakos, Athanasios V. Physics Division National Center for Theoretical Sciences National Taiwan University Taipei 10617 Taiwan Department of Computer Science and Information Engineering National Kaohsiung University of Science and Technology Kaohsiung 807618 Taiwan Center for AI Research (CAIR) University of Agder (UiA) Grimstad Norway

A dominating set of a graph G= (V, E) is a subset U of its vertices V, such that any vertex of G is either in U, or has a neighbor in U. The dominating-set problem is to find a minimum dominating set in G. Dominating sets are of critical importance for various types of networks/graphs, and find therefore potential applications in many fields. Particularly, in the area of communication, dominating sets are prominently used in the efficient organization of large-scale wireless ad hoc and sensor networks. However, the dominating set problem is also a hard optimization problem and thus currently is not efficiently solvable on classical computers. Here, we propose a biomolecular and a quantum algorithm for this problem, where the quantum algorithm provides a quadratic speedup over any classical algorithm. We show that the dominating set problem can be solved in O(2 ⁿ^/²) queries by our proposed quantum algorithm, where n is the number of vertices in G. We also demonstrate that our quantum algorithm is the best known procedure to date for this problem. We confirm the correctness of our algorithm by executing it on IBM Quantum’s qasm simulator and the Brooklyn superconducting quantum device. And lastly, we show that molecular solutions obtained from solving the dominating set problem are represented in terms of a unit vector in a finite-dimensional Hilbert space. © 2023, The Author(s).

关键词：

Solar panels salvaging for safety & sustainability

学校读者我要写书评

暂无评论

Solar panels salvaging for safety & sustainability

作者： Rathi, Kartik Vyas, Bhavesh Acharya, Puja Vyas, Jayesh Dixit, Anil Department of Computer Science AI&ML SOET K R Mangalam University Haryana Gurgaon122103 India Department of EEE SOET K R Mangalam University Haryana Gurgaon122103 India Department of ECE SOET K R Mangalam University Haryana Gurgaon122103 India Department of Mechanical and Chemical Engineering Indian Institute of Technology Jammu India Division of Research and Innovation Uttaranchal University Dehradun248007 India

This paper deals with issue causing tsunami of solar panels to be fixed in existing sustainable energy market. Raising a concern in the direction of recycling and minerals extraction. With the inclination over utilizing green energy ways of power production. The solar sector has been ambushed and still in progress to satisfy the daily surpassing limits of power demand. Being a better choice to reduce carbon emission the solar setups are in a process of capturing acres of lands and water bodies. As a result, all along with it, the life span or service period of solar panel also needs to be on count. Off servicing a solar panel assembly leaves its set up open and exposed to the environment. Its construction utilizes minerals like cadmium, gallium, germanium, tellurium etc. all in creation of modules but after its lifespan, majority of panel assemblies found at garbage's etc. and serves as pollutant. This linear economy chain needs to be channelized. High need of recycling framework at government, industry and residential level is required to provide better end of life disposal. Better termination of the minerals extracted from earth is required by fixing solar module. © 2022

关键词： Waste disposal

Anomaly Detection in Collider Physics via Factorized Observables

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Metodiev, Eric M. Thaler, Jesse Wynne, Raymond Center for Theoretical Physics Massachusetts Institute of Technology CambridgeMA02139 United States The NSF AI Institute for Artificial Intelligence and Fundamental Interactions Department of Electrical Engineering and Computer Science Massachusetts Institute of Technology CambridgeMA02139 United States Division of Physics Mathematics and Astronomy Caltech PasadenaCA United States

To maximize the discovery potential of high-energy colliders, experimental searches should be sensitive to unforeseen new physics scenarios. This goal has motivated the use of machine learning for unsupervised anomaly detection. In this paper, we introduce a new anomaly detection strategy called FORCE: factorized observables for regressing conditional expectations. Our approach is based on the inductive bias of factorization, which is the idea that the physics governing different energy scales can be treated as approximately independent. Assuming factorization holds separately for signal and background processes, the appearance of non-trivial correlations between low- and high-energy observables is a robust indicator of new physics. Under the most restrictive form of factorization, a machine-learned model trained to identify such correlations will in fact converge to the optimal new physics classifier. We test FORCE on a benchmark anomaly detection task for the Large Hadron Collider involving collimated sprays of particles called jets. By teasing out correlations between the kinematics and substructure of jets, our method can reliably extract percent-level signal fractions. This strategy for uncovering new physics adds to the growing toolbox of anomaly detection methods for collider physics with a complementary set of assumptions. Copyright © 2023, The Authors. All rights reserved.

关键词： Factorization

An Improved Time Feedforward Connections Recurrent Neural Networks

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Wang, Jin Zou, Yongsong Lim, Se-Jung School of Hydraulic & Environmental Engineering Changsha University of Science & Technology Changsha410014 China School of Computer & Communication Engineering Changsha University of Science & Technology Changsha410014 China AI Liberal Arts Studies Division of Convergence Honam University Gwangju-si62399 Korea Republic of

Recurrent Neural Networks (RNNs) have been widely applied to deal with temporal problems, such as flood forecasting and financial data processing. On the one hand, traditional RNNs models amplify the gradient issue due to the strict time serial dependency, making it difficult to realize a long-term memory function. On the other hand, RNNs cells are highly complex, which will significantly increase computational complexity and cause waste of computational resources during model training. In this paper, an improved Time Feedforward Connections Recurrent Neural Networks (TFC-RNNs) model was first proposed to address the gradient issue. A parallel branch was introduced for the hidden state at time t-2 to be directly transferred to time t without the nonlinear transformation at time t-1. This is effective in improving the long-term dependence of RNNs. Then, a novel cell structure named Single Gate Recurrent Unit (SGRU) was presented. This cell structure can reduce the number of parameters for RNNs cell, consequently reducing the computational complexity. Next, applying SGRU to TFC-RNNs as a new TFC-SGRU model solves the above two difficulties. Finally, the performance of our proposed TFC-SGRU was verified through several experiments in terms of long-term memory and anti-interference capabilities. Experimental results demonstrated that our proposed TFC-SGRU model can capture helpful information with time step 1500 and effectively filter out the noise. The TFC-SGRU model accuracy is better than the LSTM and GRU models regarding language processing ability. © 2022, CC BY.

关键词： Long short-term memory