检索结果-内蒙古大学图书馆

Latent Lab: Large Language Models for Knowledge Exploration

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Dunnell, Kevin Painter, Trudy Stoddard, Andrew Lippman, Andy Department of Electrical Engineering and Computer Science MIT United States MIT Media Lab United States

This paper investigates the potential of AI models, particularly large language models (LLMs), to support knowledge exploration and augment human creativity during ideation. We present "Latent Lab" an interactive tool for discovering connections among mit Media Lab research projects, emphasizing "exploration" over search. The work offers insights into collaborative AI systems by addressing the challenges of organizing, searching, and synthesizing content. In a user study, the tool's success was evaluated based on its ability to introduce users to an unfamiliar knowledge base, ultimately setting the groundwork for the ongoing advancement of human-AI knowledge exploration systems. © 2023, CC BY-NC-SA.

关键词： Knowledge based systems

Secretary Problems with Random Number of Candidates: How Prior Distributional Information Helps

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Zhang, Junhui Jaillet, Patrick ORC MIT United States ORC Department of Electrical Engineering and Computer Science MIT United States

We study variants of the secretary problem, where N, the number of candidates, is a random variable, and the decision maker wants to maximize the probability of success – picking the largest number among the N candidates – using only the relative ranks of the candidates revealed so far. We consider three forms of prior information about p, the probability distribution of N . In the full information setting, we assume p to be fully known. In that case, we show that single-threshold type of strategies can achieve 1/e-approximation to the maximum probability of success among all possible strategies. In the upper bound setting, we assume that N ≤ n (or E[N ] ≤ µ), where n (or µ) is known. In that case, we show that randomization over single-threshold type of strategies can achieve the optimal worst case probability of success of log(1n) (or log(1µ) ) asymptotically. Surprisingly, there is a single-threshold strategy (depending on n) that can succeed with probability 2/e2 for all but an exponentially small fraction of distributions supported on [n]. In the sampling setting, we assume that we have access to m samples N (1), . . ., N (m) ∼iid p. In that case, we show that if N ≤ T with probability at least 1 − O(ǫ) for some T ∈ N, m & ǫ12 max(log(1ǫ ), ǫ log(log(ǫT) )) is enough to learn a strategy that is at least ǫ-suboptimal, and we provide a lower bound of Ω(ǫ12 ), showing that the sampling algorithm is optimal when ǫ = O(log log(1T) ). We also discuss applications of some of our techniques to the matroid secretary problems, iid prophet inequalities, and online linear optimization problems when the time horizon N is a random variable. © 2023, CC BY.

关键词： Random variables

Digital Marketing Strategy and Consumer Behavior Analysis Through Deep Q-Network (DQN) Models 4

学校读者我要写书评

暂无评论

Digital Marketing Strategy and Consumer Behavior Analysis Th...

4th IEEE International Conference on Mobile Networks and Wireless Communications, ICMNWC 2024

作者： Shaikh, Irfan Abdul Karim Devi, K Abirami Deshmukh, Pooja Rajaram, Mallothu Unni, Manu Vasudevan Bhatt, Arti School of Business Bahrain Polytechnic Isa Town Bahrain Bhaktavatsalam Memorial College for Women Department of Commerce Chennai India Mit Group of Institutions Department of Management Science Chhatrapati Sambhajinagar India St. Martin's Engineering College Department of Computer Science and Engineering Dhulapally India Department of Management Bangalore India Shri Guru Ram Rai University Department of Mass Communication Dehradun India

ISBN: (纸本)9798350352931

Understanding digital client behaviour is essential for marketing optimisation in today's corporate environment. This study examines how online information searches, social media involvement, e-commerce buying patterns, and content consumption affect consumer behaviour. The study examines how data-driven decision-making, targeted social media advertising, tailored content, and enhanced e-commerce platforms can boost marketing. The suggested strategy involves data preparation and model training. The study highlights choosing the optimal data preprocessing methods to improve model performance. A Double Deep Q-Network (DDQN) model predicted customer behaviour after data preprocessing. The DDQN model outperformed ANN and SVM models with 93.07% accuracy. This study underlines how complex neural networks improve consumer behaviour model predicting. The findings highlight the necessity for tailored digital strategies to boost marketing efficiency and engagement across digital platforms. © 2024 IEEE.

关键词： Marketplaces

A New Switched Reluctance Motor with Embedded Permanent Magnets for Transportation Electrification

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Davarpanah, Gholamreza Mohammadi, Sajjad Department of Electrical Engineering Amirkabir University of Technology Tehran Iran Department of Electrical Engineering and Computer Science MIT Cambridge United States

A new three-phase hybrid-excited multi-tooth switched reluctance motor with embedded permanent magnets is proposed, capable of achieving higher torque density for transportation electrification applications. Operating principles and design considerations are discussed. A magnetic equivalent circuit is developed. Finite element method is employed in the field analysis. The advantages of the proposed topology over existing designs for switched reluctance motors and flux switching motors are presented. Finally, the optimized design is prototyped to experimentally confirm the design and simulation results. Copyright © 2024, The Authors. All rights reserved.

关键词： Permanent magnets

IoT Based Monitoring and Controlling Mechanism for Orange Orchard

学校读者我要写书评

暂无评论

IoT Based Monitoring and Controlling Mechanism for Orange Or...

Emerging Trends in engineering and Medical sciences (ICETEMS), International Conference on

作者： Pratibha Waghale Lalit Damahe Swati Gaikwad Chanchala Tripathi Shilpa Dhopte Amruta Paul Department of Computer Science and Engineering Yeshwantrao Chavan College of Engineering Nagpur India Department of pharmaceutics Nagpur College of Pharmacy Wanadongri Nagpur India Dept. of Computer Science and Engineering MIT School of Computing MIT Art Designand Technology University Pune India

ISBN: (数字)9798331529352

ISBN: (纸本)9798331529390

In orange fields, precision agriculture is essential for farmer prosperity, yet challenges like soil fertility persist. This study explores implementation of IoT (Internet of Things) technology for monitoring orange cultivation in Nagpur's agricultural landscape. With the economic prosperity of local farmers intricately linked to the management of agronomic variables, the inherent uncertainties surrounding factors like soil fertility andenvironmental dynamics pose formidable challenges. Thisproject presents an autonomous sensor interface for an 1oT-based monitoring and control system for orange orchards. The real working prototype consisting of Soil moisture sensor, DHT11 sensor to measure humidity and NPK sensors is designed and implemented. The Blynk mobile application and OLED display provides real-time monitoring capabilities. The Thingspeak platform helps asthe main base station for the online storage of sensorreadings. A dedicated mobile application is also provided to the farmers to get insights about soil nutrient levels and fertilizer requirement. By harnessing the transformative potential of IoT technology, this precision agriculture framework represents a significant advancement in the cultivation of oranges, promising enhanced economic prosperity and sustainability for Nagpur's farming community.

关键词： Temperature measurement Temperature sensors Soil measurements Soil moisture Organic light emitting diodes Nutrients Real-time systems Mobile applications Internet of Things Monitoring

Comparative Analysis of Economic Power Dispatch with PSO and IPSO for Consumer Load System

学校读者我要写书评

暂无评论

Comparative Analysis of Economic Power Dispatch with PSO and...

2024 International Conference on Sustainable Computing and Integrated Communication in Changing Landscape of AI, ICSCAI 2024

作者： Verma, Aman Chinthamu, Narender Shnain, Ammar Hameed Nainwal, Ankita Badhoutiya, Arti Gupta, Priyanka Upadhyay, Shrikant Lloyd Institute of Engineering & Technology Greater Noida India MIT CTO DallasTX United States Department of Computers Techniques Engineering College of Technical Engineering The Islamic University Najaf Iraq Department of Computer Science & Engineering Graphic Era Deemed to be University Dehradun India Department of Electrical Engineering GLA University Mathura India Institute of Aeronautical Engineering Dundigal Hyderabad India Department of Electronics and Communication Engineering MLR Institute of Technology Hyderabad India

ISBN: (纸本)9798350355338

In the power industry, the generating systems' financial viability is always crucial. That is one of the many problems with the electrical grid. Choosing the optimal generator pairing to lower system operating costs should be the goal of the financial power dispatch, taking into consideration variations in the average demand pattern over a specific amount of time. In order to solve the DELD issue in the most effective way possible, this will use the recommended solution approach based on PSO Strategy in an effort to lower operational expenses and fuel consumption. On a system with five producing structures, the simulation is run employing the process of iteration optimum course of a 24-hour period, and the results are compared to those obtained using the normal PSO and the rest of the traditional approaches. © 2024 IEEE.

关键词： Operating costs

Tamper-forgery Image Detection in Security Digital Image: A Review

学校读者我要写书评

暂无评论

Tamper-forgery Image Detection in Security Digital Image: A ...

Blockchain and Distributed Systems Security (ICBDS), IEEE International Conference on

作者： Nilesh N. Thorat Prashant Dhotre Nilesh Kulal Aarti P Pimpalkar Vijaya S. Patil Sachin M. Kolekar Computer Science & Engineering Department MIT School of Computing MIT Art Design and Technology University Pune Information Technology Department MIT School of Computing MIT Art Design and Technology University Pune

ISBN: (数字)9798350354348

ISBN: (纸本)9798350354355

Concerns with image security exist in every sector that uses digital photographs. Historically, public safety and forensics have depended on images from the crime scene, biometric photos, suspect photos, and other sources. With the development of digital imaging, there has been a significant increase in the use of digital images in this field. Although the use of digital image processing has led to the creation of several new forensic methods of inquiry, it has also simplified picture modification. Digital picture validity is an issue since several snipping and image alteration programs are freely available to the public. It serves as solid proof in many different types of crimes and is utilized as documentation for many purposes. Creating and modifying photos has become easier and more accessible with the development of image manipulation and editing tools. The two most common methods of photo counterfeiting are copy-and-move forgery and image splicing. An image can be spliced, or combined from two photos, to hide or show an error scenario. A section of a picture can be copied and pasted into another image. The many kinds of digital image frauds are examined in this study along with software that detects them. An analysis was done on the methods currently in use for spotting phony photos.

关键词： Reviews Digital images Forensics Splicing Training data Solids Forgery Software Fraud Security

IMPROVING WORLD MODELS USING DEEP SUPERVISION WITH LINEAR PROBES

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Zahorodnii, Andrii Department of Electrical Engineering and Computer Science MIT United States Department of Brain and Cognitive Sciences McGovern Institute for Brain Research MIT CambridgeMA United States

Developing effective world models is crucial for creating artificial agents that can reason about and navigate complex environments. In this paper, we investigate a deep supervision technique for encouraging the development of a world model in a network trained end-to-end to predict the next observation. While deep supervision has been widely applied for task-specific learning, our focus is on improving the world models. Using an experimental environment based on the Flappy Bird game, where the agent receives only LIDAR measurements as observations, we explore the effect of adding a linear probe component to the network’s loss function. This additional term encourages the network to encode a subset of the true underlying world features into its hidden state. Our experiments demonstrate that this supervision technique improves both training and test performance, enhances training stability, and results in more easily decodable world features – even for those world features which were not included in the training. Furthermore, we observe a reduced distribution drift in networks trained with the linear probe, particularly during high-variability phases of the game (flying between successive pipe encounters). Including the world features loss component roughly corresponded to doubling the model size, suggesting that the linear probe technique is particularly beneficial in compute-limited settings or when aiming to achieve the best performance with smaller models. These findings contribute to our understanding of how to develop more robust and sophisticated world models in artificial agents, paving the way for further advancements in this field. Copyright © 2025, The Authors. All rights reserved.

关键词：

Hybrid CNN and Random Forest Model with Late Fusion for Detection of Autism Spectrum Disorder in Toddlers

学校读者我要写书评

暂无评论

SSRN

SSRN 2025年

作者： Nawghare, Pushpmala Sampatrao Prasad, Jayashree Department of Computer Science and Engineering MIT School of Computing MIT Art Design and Technology University Maharashtra Pune412201 India Sinhgad College of Engineering Vadgaon Pune India

Accurate and early diagnosis of Autism Spectrum Disorder (ASD) in toddlers is crucial for effective intervention. Traditional models like Support Vector Machines (SVM), Logistic Regression (LR), and Random Forest (RF) have shown limited success, with accuracies around 65%, while deep learning approaches such as transfer learning with Convolutional Neural Networks (CNNs) achieve higher accuracies, ranging from 75% to 85%. Building on this progress, our study proposes a hybrid model combining VGG16, a pre-trained deep CNN, with an RF classifier to leverage high-level image feature extraction using the ACD image dataset on Kaggle alongside robust decision-making on the ACD Questionnaire dataset. The proposed model achieves an accuracy of 88.34%, outperforming both standalone deep learning models like VGG16, EfficientNetB0, and AlexNet-based models as well as conventional ML models. This improvement demonstrates the effectiveness of combining feature-rich deep learning outputs with RF's ensemble-based classification. Our findings suggest that this hybrid approach is highly suitable for ASD classification tasks, enhancing the reliability of predictions in clinical settings. This research not only establishes the model as an option for ASD diagnosis but also underscores the potential of hybrid architectures that fuse deep learning with machine learning. © 2025, The Authors. All rights reserved.

关键词： Convolutional neural networks