Summary of Neural Network Approximators For Marginal Map in Probabilistic Circuits, by Shivvrat Arya et al.
Neural Network Approximators for Marginal MAP in Probabilistic Circuits
by Shivvrat Arya, Tahrima Rahman, Vibhav Gogate
First submitted to arxiv on: 6 Feb 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI)
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 In this paper, researchers propose a novel approach to solving maximum-a-posteriori (MAP) and marginal MAP (MMAP) tasks in probabilistic circuits (PCs). PCs, such as sum-product networks, efficiently represent large multi-variate probability distributions and are preferred over other representations like Bayesian and Markov networks due to their linear time complexity for marginal inference. However, the MAP and MMAP tasks remain NP-hard in these models. The authors draw inspiration from recent work on neural networks for optimization problems and develop a method that uses neural networks to approximate (M)MAP inference in PCs. This approach is self-supervised and requires only linear time to output a solution after the neural network is learned. The proposed method outperforms three competing linear-time approximations, max-product inference, max-marginal inference, and sequential estimation. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary In this paper, scientists found a way to solve complex problems in computers called probabilistic circuits (PCs). PCs are used to understand large amounts of data and make predictions. The problem is that it takes a long time to get the best answer for these questions. The researchers looked at how neural networks can be used to help with this problem. They created a new method that uses neural networks to find good answers quickly, without needing to look at all the information. This approach works well and is faster than other methods used today. |
Keywords
* Artificial intelligence * Inference * Neural network * Optimization * Probability * Self supervised
