Summary of Initial Investigation Of Kolmogorov-arnold Networks (kans) As Feature Extractors For Imu Based Human Activity Recognition, by Mengxi Liu et al.
Initial Investigation of Kolmogorov-Arnold Networks (KANs) as Feature Extractors for IMU Based Human Activity Recognition
by Mengxi Liu, Daniel Geißler, Dominique Nshimyimana, Sizhen Bian, Bo Zhou, Paul Lukowicz
First submitted to arxiv on: 16 Jun 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Emerging Technologies (cs.ET); Signal Processing (eess.SP)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper explores using a novel neural network architecture, Kolmogorov-Arnold Networks (KANs), as feature extractors for sensor-based Human Activity Recognition (HAR) tasks. Unlike conventional networks, KANs perform non-linear computations represented by B-SPLINES on the edges leading to each node and then sum up the inputs at the node. This allows the network to learn spline parameters instead of weights. The paper hypothesizes that this ability is advantageous for computing low-level features in IMU-based HAR tasks. Four architecture variations are implemented, and an initial performance investigation shows that KAN-based feature extractors outperform CNN-based extractors on all datasets while being more parameter efficient. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper uses a new kind of neural network to help computers recognize human activities from sensor data. This type of network is called Kolmogorov-Arnold Networks (KANs). The researchers think that KANs can be better at finding patterns in the sensor data than other types of networks. They test this idea by using KANs to recognize different human activities from four sets of data. The results show that KANs are better than other networks and use fewer “building blocks” (called parameters) to get those results. |
Keywords
» Artificial intelligence » Activity recognition » Cnn » Neural network » Parameter efficient
