This MATLAB script presents an innovative approach to 5G beamforming prediction using a sequence-based LSTM neural network. Unlike conventional methods that predict only final vectors, this solution provides time-stepped predictions across entire sequences, enabling real-time tracking of dynamic channel conditions. The framework achieves stable training convergence while maintaining physically meaningful performance metrics, including realistic path loss and SNR values.
The data was collected by a tester holding a Xiaomi 13 smartphone while walking and collecting data in an underground parking lot covering a 16x70m area. The data includes 5G radio features and geomagnetic field information.
Collection Time: From 09:58 AM to 10:34 AM on July 13, 2024.
Total Samples: 12,800
Training Set (including validation set): 10,240
Test Set: 2,560
This paper conducts in-depth research on three text classification tasks: sentiment analysis, offensive language identification, and news topic classification.
In this dataset, a human detecting model using with UWB radar technology is presented. Two distinct datasets were created using the UWB radar device, leveraging its dual features. Data collection involved two main scenarios, each containing multiple sub-scenarios. These sub-scenarios varied parameters like the position, distance, angle, and orientation of the human subject relative to the radar. Unlike conventional approaches that rely on signal processing or noise/background removal, this study uniquely emphasizes analyzing raw UWB radar data directly.
The JU-Impact Radiomap Dataset is a comprehensive dataset designed for research and development in indoor positioning systems. It comprises 5431 instances characterized by readings from 105 static Wi-Fi Access Points (APs) and spans 152 distinct virtual grids. Each virtual grid represents a 1x1 square meter area, derived by dividing a physical floor of a university building into reference coordinate points (x, y). The dataset was collected over a period of 21 days using four mobile devices: Samsung Galaxy Tab, Moto G, Redmi Note 4, and Google Pixel.
FLAME 3 is the third dataset in the FLAME series of aerial UAV-collected side-by-side multi-spectral wildlands fire imagery (see FLAME 1 and FLAME 2). This set contains a single-burn subset of the larger FLAME 3 dataset focusing specifically on Computer Vision tasks such as fire detection and segmentation.
Dateset of the Three-Phase Flow Facility. The Three-phase Flow Facility at Cranfield University is designed to provide a controlled and measured flow rate of water, oil and air to a pressurized system. Fig. 1 shows a simplified sketch of the facility. The test area consists of pipelines with different bore sizes and geometries, and a gas and liquid two-phase separator (0.5 m diameter and 1.2 m high) at the top of a 10.5 m high platform. It can be supplied with single phase of air, water and oil, or a mixture of those fluids, at required rates.
We propose MM-Vet v2, an evaluation benchmark that examines large multimodal models (LMMs) on complicated multimodal tasks. Recent LMMs have shown various intriguing abilities, such as solving math problems written on the blackboard, reasoning about events and celebrities in news images, and explaining visual jokes. Rapid model advancements pose challenges to evaluation benchmark development.
This dataset is specifically designed for smart contract code search, aiming to facilitate research and development in this domain. To construct the dataset, we collected annotated Solidity smart contract code from two major sources: GitHub and Etherscan. In total, we compiled approximately 470K raw (code, docstring) pairs, ensuring a diverse and comprehensive dataset.
Dataset 1 include 100 rebar-reinforced rectangular UHPC beams data.
