---
title: Generalizability of Surrogate Models
url: https://www.ml-quant.com/papers/ssrn/5013717/
site: ML-Quant (https://www.ml-quant.com)
updated: 2026-09-26
license: Summaries CC BY 4.0; links go to the original sources
index: https://www.ml-quant.com/llms.txt
identifier: SSRN 5013717
source_url: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5013717
featured: 2024-11-13
citations: unknown
topic: ML & AI Methods
---


# Generalizability of Surrogate Models

A study highlights the potential of deep learning models to extract input data from output data in building energy modelling.

- Source: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5013717
- Identifier: SSRN 5013717
- Released: 2024-11-08
- First featured: Quant Letter No. 74 (2024-11-13): https://www.ml-quant.com/issues/2024-11-13/
- Citations (Semantic Scholar): not tracked
- Published in: not yet
- Topic: ML & AI Methods

## Related

- [Transformers are SSMs: Generalized Models and Efficient Algorithms Through Structured State Space Duality](https://www.ml-quant.com/papers/arxiv/2405.21060/): The research identifies a link between state-space models and Transformers in deep learning, leading to the creation of a faster language modeling architecture, Mamba-2.
- [SOAP: Improving and Stabilizing Shampoo using Adam](https://www.ml-quant.com/papers/arxiv/2409.11321/): A new algorithm, SOAP, enhances the computational efficiency of the Shampoo preconditioning method in deep learning tasks, reducing iterations and time, with an online implementation available.
- [Edge Directionality Improves Learning on Heterophilic Graphs](https://www.ml-quant.com/papers/arxiv/2305.10498/): The study presents Directed Graph Neural Network (Dir-GNN), a new deep learning framework for directed graphs that surpasses traditional models in heterophilic benchmarks.
- [TabR: Tabular Deep Learning Meets Nearest Neighbors](https://www.ml-quant.com/papers/arxiv/2307.14338/): Tabular DL Meets Nearest Neighbors: TabR, a new deep learning model for tabular data, outperforms existing models by using a k-Nearest-Neighbors-like component for better predictions.
- [Modular Duality in Deep Learning](https://www.ml-quant.com/papers/arxiv/2410.21265/): The article presents a new theory of modular dualization for general neural networks, providing a theoretical basis for fast and scalable training algorithms, potentially leading to a new generation of optimizers for neural architectures.
- [Mechanistic Design and Scaling of Hybrid Architectures](https://www.ml-quant.com/papers/arxiv/2403.17844/): The creation of deep learning architectures can be streamlined using an end-to-end design pipeline, which uses synthetic tasks to predict scaling laws and find optimal architectures.