Machine Learning

Surface electromyography (EMG) can be used to interact with and control robots via intent recognition. However, most machine learning algorithms used to decode EMG signals have been trained on small datasets with limited subjects, impacting their generalization across different users and tasks. Here we developed EMGNet, a large-scale dataset for EMG neural decoding of human movements. EMGNet combines 7 open-source datasets with processed EMG signals for 132 healthy subjects (152 GB total size).

This dataset presents real-world VPN encrypted traffic flows captured from five applications that belong to four service categories, which reflect typical usage patterns encountered by everyday users. 

Our methodology utilized a set of automatic scripts to simulate real-world user interactions for these applications, to achieve a low level of noise and irrelevant network traffic.

The dataset consists of flow data collected from four service categories:

DALHOUSIE NIMS LAB BENIGN DATASET 2024-2 dataset comprises data captured from Consumer IoT devices, depicting three primary real-life states (Power-up, Idle, and Active) experienced by everyday users. Our setup focuses on capturing realistic data through these states, providing a comprehensive understanding of Consumer IoT devices.

The dataset comprises of nine popular IoT devices namely 

Amcrest Camera

Smarter Coffeemaker

Ring Doorbell

Amazon Echodot

Google Nestcam

Google Nestmini

Kasa Powerstrip

To download the dataset without purchasing an IEEE Dataport subscription, please visit: https://zenodo.org/records/13738598

Please cite the following paper when using this dataset:

N. Thakur, “Mpox narrative on Instagram: A labeled multilingual dataset of Instagram posts on mpox for sentiment, hate speech, and anxiety analysis,” arXiv [cs.LG], 2024, URL: https://arxiv.org/abs/2409.05292

Abstract

5G cellular networks are particularly vulnerable against narrowband jammers that target specific control subchannels in the radio signal. One mitigation approach is to detect such jamming attacks with an online observation system, based on machine learning. We propose to detect jamming at the physical layer with a pre-trained machine-learning model that performs binary classification. Based on data from an experimental 5G network, we study the performance of different classification models.

This dataset contains 570 JPEG images of electricity meters taken from varied locations within the IIT BHU campus, including the GTFRC and residential apartments. It showcases a broad range of real-world scenarios, with each image demonstrating different challenges such as varying lighting conditions, levels of focus and clarity, and a wide range of capture angles. These attributes test and enhance the robustness of technologies designed to interpret meter readings from photographs under diverse conditions.

Measuring and assessing intelligence level in children and adolescents is crucial for monitoring their developmental progress, identifying intellectual disabilities, and implementing early interventions. To date, there is no digital and simplified tool specifically designed to evaluate whether intelligence is normal or abnormal in these age stages.

Recently, combinatorial interaction strategies have a large spectrum as black box strategies for testing software and hardware. This paper discusses a novel adoption of a combinatorial interaction strategy to generate a sparse combinatorial data table (SCDT) for machine learning. Unlike test data generation strategies, in which the t-way tuples synthesize into a test case, the proposed SCDT requires analyzing instances against their corresponding tuples to generate a systematic learning dataset.