A novel multiple instance learning method based on extreme learning machine.

Area of Science:

• Machine Learning
• Artificial Intelligence

Background:

• Real-world datasets often contain numerous instances, necessitating efficient algorithms.
• Multiple Instance Learning (MIL) is a machine learning paradigm distinct from supervised learning, dealing with bags of unlabeled instances.

Purpose of the Study:

• To propose a novel and efficient algorithm for Multiple Instance Learning (MIL).
• To address the challenge of large datasets in MIL by developing a faster and effective method.

Main Methods:

• A novel method based on Extreme Learning Machine (ELM) is proposed for MIL.
• A single hidden layer feedforward network (SLFN) with random weights selects the most qualified instance from each bag.
• The selected instance represents the bag, and a modified ELM model is trained using these instances.

Main Results:

• The proposed ELM-MIL algorithm demonstrates good performance on benchmark datasets.
• Experiments include both classification and regression tasks within the MIL framework.
• The ELM-MIL algorithm achieves significant speed improvements, running several to hundreds of times faster than comparable MIL algorithms.

Conclusions:

• The ELM-MIL algorithm offers an efficient and effective solution for Multiple Instance Learning problems.
• This approach is particularly beneficial for handling large-scale real-world datasets.
• The method provides a substantial performance advantage in terms of computational speed.