h4cker/ai_research/ethics_privacy/README.md
2023-09-08 11:54:12 -04:00

92 lines
7.1 KiB
Markdown

# AI Ethics and Privacy Resources
### Databases for Human Activity Recognition:
1. **[MobiAct](https://github.com/MatheLi/Fall_Detection_App_AI/blob/master/posts/The_dataset.md)**
A dataset optimized for detecting activities such as falls, walking, and jogging. It is primarily used in creating apps that use smartphone sensors to detect falls, particularly in elderly individuals.
2. **[NHANES Dataset](http://www.sal.disco.unimib.it/technologies/unimib-shar/)**
Although not exclusively designed for HAR, the NHANES dataset is a rich source of health and nutritional data, which could potentially be utilized to garner insights into human activities and health conditions.
3. **[UniMiB SHAR](https://wwwn.cdc.gov/nchs/nhanes/)**
This repository houses data concerning human activities collected from smartphone accelerometer sensors. It serves as a valuable resource for developing machine learning models capable of recognizing various activities.
4. **[UCI Human Activity Recognition Using Smartphones Dataset](https://archive.ics.uci.edu/dataset/240/human+activity+recognition+using+smartphones)**
This dataset comprises data from smartphone accelerometers and gyroscopes, capturing activities such as walking, sitting, and standing performed by 30 subjects. It is a popular choice for HAR research projects.
5. **[ISDM (Wireless Sensor Data Mining)](https://github.com/topics/wireless-sensor-data-mining)**
Although not a database per se, this GitHub topic connects you to various resources and datasets pertaining to wireless sensor data mining, an essential aspect in HAR research.
6. **[HHAR (Heterogeneity Human Activity Recognition)](https://github.com/Limmen/Distributed_ML)**
HHAR stands out with its data collected from a range of devices, portraying various human activities. It is particularly beneficial for constructing models adaptable to different data sources.
7. **[PAMAP2 Physical Activity Monitoring](https://archive.ics.uci.edu/dataset/231/pamap2+physical+activity+monitoring)**
Featuring data from wearable sensors monitoring individuals performing diverse physical activities, PAMAP2 is a vital tool for developing predictive HAR models.
8. **[Daphnet Freezing of Gait](https://archive.ics.uci.edu/dataset/245/daphnet+freezing+of+gait)**
Focused on Parkinson's patients' gait freezing, this dataset, comprising data from wearable sensors, plays a crucial role in HAR healthcare applications.
9. **[Actitracker](https://github.com/gomahajan/har-actitracker)**
Developed to recognize various physical activities through smartphone sensors, Actitracker houses data on activities such as walking and jogging.
10. **[Daily and Sports Activities](https://archive.ics.uci.edu/dataset/256/daily+and+sports+activities)**
This dataset contains data on a range of daily and sports activities recorded through wearable sensors, making it a rich resource for HAR research, especially in distinguishing between different physical activities.
11. **[Smartphone Dataset for Human Activity Recognition (HAR) in Ambient Assisted Living (AAL)](https://archive.ics.uci.edu/dataset/364/smartphone+dataset+for+human+activity+recognition+har+in+ambient+assisted+living+aal)**
This dataset focuses on aiding the elderly or disabled, using smartphone sensors to identify their activities, hence fostering safer and more comfortable living environments.
12. **[Opportunity Activity Recognition](https://archive.ics.uci.edu/dataset/226/opportunity+activity+recognition)**
This dataset is notable for its emphasis on context recognition, using sensor data from various sources to identify complex activities and gestures, thereby advancing research in ambient intelligence.
13. **[CASAS](https://casas.wsu.edu/datasets/)**
CASAS, a collection of datasets centered on smart home environments, facilitates the creation of algorithms capable of recognizing home-based activities through sensor data.
14. **[MSR Daily Activity 3D](https://wangjiangb.github.io/my_data.html)**
This dataset distinguishes itself with its inclusion of depth maps alongside skeletal data for activity recognition, aiding in the development of models capable of identifying activities from 3D data.
15. **[REALDISP Activity Recognition Dataset](https://mldta.com/dataset/realdisp-activity-recognition-dataset/)**
REALDISP incorporates data on various activities captured through wearable sensors, with a focus on realistic data disposition, which is vital for creating robust HAR models.
### Tools & Methods for Data Collection, Cleaning, and Analysis:
- **Data Collection**:
- APIs and SDKs
- Wireless transmission
### Data Cleaning:
3. **Pandas**:
- **Example**: Cleaning a dataset with missing values using Pandas before training a machine learning model.
- **Relevant Link**: [Pandas Documentation](https://pandas.pydata.org/pandas-docs/stable/index.html)
- **Usage in HAR and AI**: Pandas can be used to structure and clean sensor data, making it suitable for training AI models capable of recognizing complex patterns in human activity data.
4. **Sci-kit learn**:
- **Example**: Using Sci-kit learn for feature selection and removing irrelevant features from a dataset.
- **Relevant Link**: [Sci-kit learn Documentation](https://scikit-learn.org/stable/)
- **Usage in HAR and AI**: Sci-kit learn offers various tools for data preprocessing, which is a vital step in preparing data for AI algorithms, enhancing the performance of the models in HAR applications.
### Data Analysis:
5. **TensorFlow**:
- **Example**: Developing a deep learning model using TensorFlow to classify different activities based on sensor data.
- **Relevant Link**: [TensorFlow Documentation](https://www.tensorflow.org/learn)
- **Usage in HAR and AI**: TensorFlow provides a comprehensive platform for developing and training AI models capable of analyzing and recognizing patterns in human activity data.
6. **Keras**:
- **Example**: Using Keras to create a convolutional neural network (CNN) for image recognition, an essential task in AI.
- **Relevant Link**: [Keras Documentation](https://keras.io/getting_started/intro_to_keras_for_engineers/)
- **Usage in HAR and AI**: Keras simplifies the process of building and optimizing neural networks, a crucial component in AI, to analyze human activity data more effectively and make predictions.
### Visualization and Further Analysis:
7. **Matplotlib**:
- **Example**: Using Matplotlib to visualize the distribution of different activities within a dataset.
- **Relevant Link**: [Matplotlib Documentation](https://matplotlib.org/stable/contents.html)
- **Usage in HAR and AI**: Visualization of data is essential in AI to understand underlying patterns and trends in data, aiding in the better development and tuning of models for HAR.
8. **Seaborn**:
- **Example**: Creating a heatmap using Seaborn to visualize the correlation between different features in a dataset.
- **Relevant Link**: [Seaborn Documentation](https://seaborn.pydata.org/)
- **Usage in HAR and AI**: Seaborn can enhance data visualization in AI, assisting in identifying relationships and patterns in data which can influence the development and performance of HAR models.