Picture for Matt Thomson

Matt Thomson

Generating counterfactual explanations of tumor spatial proteomes to discover effective, combinatorial therapies that enhance cancer immunotherapy

Nov 08, 2022
Zitong Jerry Wang, Matt Thomson

Figure 1 for Generating counterfactual explanations of tumor spatial proteomes to discover effective, combinatorial therapies that enhance cancer immunotherapy

Figure 2 for Generating counterfactual explanations of tumor spatial proteomes to discover effective, combinatorial therapies that enhance cancer immunotherapy

Figure 3 for Generating counterfactual explanations of tumor spatial proteomes to discover effective, combinatorial therapies that enhance cancer immunotherapy

Figure 4 for Generating counterfactual explanations of tumor spatial proteomes to discover effective, combinatorial therapies that enhance cancer immunotherapy

Access Paper or Ask Questions

Engineering flexible machine learning systems by traversing functionally invariant paths in weight space

May 09, 2022
Guruprasad Raghavan, Matt Thomson

Figure 1 for Engineering flexible machine learning systems by traversing functionally invariant paths in weight space

Figure 2 for Engineering flexible machine learning systems by traversing functionally invariant paths in weight space

Figure 3 for Engineering flexible machine learning systems by traversing functionally invariant paths in weight space

Figure 4 for Engineering flexible machine learning systems by traversing functionally invariant paths in weight space

* 17 pages

Access Paper or Ask Questions

Active feature selection discovers minimal gene-sets for classifying cell-types and disease states in single-cell mRNA-seq data

Jun 15, 2021
Xiaoqiao Chen, Sisi Chen, Matt Thomson

* 19 pages, 4 figures

Access Paper or Ask Questions

Solving hybrid machine learning tasks by traversing weight space geodesics

Jun 05, 2021
Guruprasad Raghavan, Matt Thomson

Figure 1 for Solving hybrid machine learning tasks by traversing weight space geodesics

Figure 2 for Solving hybrid machine learning tasks by traversing weight space geodesics

Figure 3 for Solving hybrid machine learning tasks by traversing weight space geodesics

Figure 4 for Solving hybrid machine learning tasks by traversing weight space geodesics

* 11 pages, 7 figures

Access Paper or Ask Questions

Sparsifying networks by traversing Geodesics

Dec 12, 2020
Guruprasad Raghavan, Matt Thomson

Figure 1 for Sparsifying networks by traversing Geodesics

Figure 2 for Sparsifying networks by traversing Geodesics

* 5 pages; Presented work at NeurIPS 2020 Workshop (DiffGeo4DL). arXiv admin note: text overlap with arXiv:2005.11603

Access Paper or Ask Questions

Self-organization of multi-layer spiking neural networks

Jun 12, 2020
Guruprasad Raghavan, Cong Lin, Matt Thomson

Figure 1 for Self-organization of multi-layer spiking neural networks

Figure 2 for Self-organization of multi-layer spiking neural networks

Figure 3 for Self-organization of multi-layer spiking neural networks

Figure 4 for Self-organization of multi-layer spiking neural networks

* 11 pages, 4 figures

Access Paper or Ask Questions

Geometric algorithms for predicting resilience and recovering damage in neural networks

Jun 02, 2020
Guruprasad Raghavan, Jiayi Li, Matt Thomson

Figure 1 for Geometric algorithms for predicting resilience and recovering damage in neural networks

Figure 2 for Geometric algorithms for predicting resilience and recovering damage in neural networks

Figure 3 for Geometric algorithms for predicting resilience and recovering damage in neural networks

Figure 4 for Geometric algorithms for predicting resilience and recovering damage in neural networks

* 10 pages and 4 figures

Access Paper or Ask Questions