# jindongwang/transferlearning architecture

> Architecture diagram of the GitHub repository jindongwang/transferlearning, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-10-07). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/jindongwang/transferlearning
- Repository: https://github.com/jindongwang/transferlearning

## Overview

This repository is a curated transfer-learning resource, not one integrated application. Its code directory contains independent implementations and experiment entry points for domain adaptation, domain generalization, speech and vision transfer, classical methods, and robustness/pruning. Thus there is no single user-to-result runtime workflow; each experiment generally selects data and a method, then trains or evaluates it. The graph highlights these distinct branches and selected internal layers. Relationships are included only where the README or source index supports them; many listed implementations were not sampled, so their internal wiring is omitted.

Main entry points: [`code/deep/ReMoS/CV_adv/nc_prune/analyze_coverage.py`](https://github.com/jindongwang/transferlearning/blob/master/code/deep/ReMoS/CV_adv/nc_prune/analyze_coverage.py) (Coverage analysis), [`code/deep/CSG/a-domainbed/main.py`](https://github.com/jindongwang/transferlearning/blob/master/code/deep/CSG/a-domainbed/main.py) (CSG experiment), [`code/DeepDA/models.py`](https://github.com/jindongwang/transferlearning/blob/master/code/DeepDA/models.py) (Transfer model), [`code/DeepDA/main.py`](https://github.com/jindongwang/transferlearning/blob/master/code/DeepDA/main.py) (DeepDA runner)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_adapt["Domain adaptation"]
  node_deepda_run["DeepDA runner<br/>[main.py]"]
  node_deepda_data["Domain data<br/>[data_loader.py]"]
  node_deepda_model["Transfer model<br/>[models.py]"]
  node_deepda_losses["Transfer losses<br/>[transfer_losses.py]"]
  node_deepda_backbones["Feature backbones<br/>[backbones.py]"]
  node_daan_train["DAAN training<br/>[train.py]"]
  node_meda_train["DeepMEDA training<br/>[main.py]"]
  node_meda_model["DeepMEDA model<br/>[deep_meda.py]"]
  node_dann_train["DaNN training<br/>[main.py]"]
  node_dann_model["DaNN network<br/>[DaNN.py]"]
end

subgraph group_generalize["Generalization methods"]
  node_deepdg_train["DeepDG runner<br/>[train.py]"]
  node_deepdg_algorithms["DG algorithms"]
  node_deepdg_loaders["DG data loaders"]
  node_csg_runner["CSG experiment<br/>[main.py]"]
  node_csg_methods["CSG methods"]
  node_csg_arch["CSG architectures<br/>[cnn.py]"]
  node_csg_data["CSG data loading<br/>[data_loader.py]"]
end

subgraph group_modal["Modalities and features"]
  node_asr_adapt["Speech adaptation"]
  node_asr_cmatch["Speech matching"]
  node_clip_runner["CLIP experiments<br/>[main.py]"]
  node_clip_model["CLIP model wrapper<br/>[clip_model.py]"]
  node_clip_data["Image-text data<br/>[data_loader.py]"]
  node_feature_run["Feature extraction<br/>[main.py]"]
end

subgraph group_classical["Classical methods"]
  node_traditional["Traditional methods"]
end

subgraph group_robust["Robustness and pruning"]
  node_remos_analysis["Coverage analysis"]
  node_remos_coverage["Neuron coverage"]
  node_remos_wrapper["Model wrapper<br/>[pytorch_wrapper.py]"]
  node_remos_strategy["Neuron selection<br/>[random.py]"]
  node_remos_datasets["Vision datasets"]
  node_remos_model["ResNet models<br/>[fe_resnet.py]"]
end

node_deepda_run -->|"loads domains"| node_deepda_data
node_deepda_run -->|"builds model"| node_deepda_model
node_deepda_model -->|"uses losses"| node_deepda_losses
node_deepda_model -->|"uses backbone"| node_deepda_backbones
node_deepdg_train -->|"selects algorithm"| node_deepdg_algorithms
node_deepdg_train -->|"gets loaders"| node_deepdg_loaders
node_csg_runner -->|"loads data"| node_csg_data
node_csg_runner -->|"uses methods"| node_csg_methods
node_csg_runner -->|"uses architectures"| node_csg_arch
node_meda_train -->|"builds model"| node_meda_model
node_dann_train -->|"builds network"| node_dann_model
node_clip_runner -->|"creates wrapper"| node_clip_model
node_clip_runner -->|"loads dataset"| node_clip_data
node_remos_analysis -->|"uses coverage"| node_remos_coverage
node_remos_analysis -->|"wraps model"| node_remos_wrapper
node_remos_analysis -->|"selects datasets"| node_remos_datasets
node_remos_analysis -->|"loads network"| node_remos_model
node_remos_analysis -->|"selects neurons"| node_remos_strategy

click node_deepda_run "https://github.com/jindongwang/transferlearning/blob/master/code/DeepDA/main.py"
click node_deepda_data "https://github.com/jindongwang/transferlearning/blob/master/code/DeepDA/data_loader.py"
click node_deepda_model "https://github.com/jindongwang/transferlearning/blob/master/code/DeepDA/models.py"
click node_deepda_losses "https://github.com/jindongwang/transferlearning/blob/master/code/DeepDA/transfer_losses.py"
click node_deepda_backbones "https://github.com/jindongwang/transferlearning/blob/master/code/DeepDA/backbones.py"
click node_daan_train "https://github.com/jindongwang/transferlearning/blob/master/code/deep/DAAN/train.py"
click node_meda_train "https://github.com/jindongwang/transferlearning/blob/master/code/deep/DeepMEDA/main.py"
click node_meda_model "https://github.com/jindongwang/transferlearning/blob/master/code/deep/DeepMEDA/deep_meda.py"
click node_dann_train "https://github.com/jindongwang/transferlearning/blob/master/code/deep/DaNN/main.py"
click node_dann_model "https://github.com/jindongwang/transferlearning/blob/master/code/deep/DaNN/DaNN.py"
click node_deepdg_train "https://github.com/jindongwang/transferlearning/blob/master/code/DeepDG/train.py"
click node_deepdg_algorithms "https://github.com/jindongwang/transferlearning/tree/master/code/DeepDG/alg/algs"
click node_deepdg_loaders "https://github.com/jindongwang/transferlearning/tree/master/code/DeepDG/datautil"
click node_csg_runner "https://github.com/jindongwang/transferlearning/blob/master/code/deep/CSG/a-domainbed/main.py"
click node_csg_methods "https://github.com/jindongwang/transferlearning/tree/master/code/deep/CSG/methods"
click node_csg_arch "https://github.com/jindongwang/transferlearning/blob/master/code/deep/CSG/arch/cnn.py"
click node_csg_data "https://github.com/jindongwang/transferlearning/blob/master/code/deep/CSG/utils/preprocess/data_loader.py"
click node_asr_adapt "https://github.com/jindongwang/transferlearning/blob/master/code/ASR/Adapter/e2e_asr_adaptertransformer.py"
click node_asr_cmatch "https://github.com/jindongwang/transferlearning/blob/master/code/ASR/CMatch/e2e_asr_udatransformer.py"
click node_clip_runner "https://github.com/jindongwang/transferlearning/blob/master/code/clip/main.py"
click node_clip_model "https://github.com/jindongwang/transferlearning/blob/master/code/clip/clip_model.py"
click node_clip_data "https://github.com/jindongwang/transferlearning/blob/master/code/clip/data/data_loader.py"
click node_feature_run "https://github.com/jindongwang/transferlearning/blob/master/code/feature_extractor/for_image_data/main.py"
click node_traditional "https://github.com/jindongwang/transferlearning/tree/master/code/traditional"
click node_remos_analysis "https://github.com/jindongwang/transferlearning/blob/master/code/deep/ReMoS/CV_adv/nc_prune/analyze_coverage.py"
click node_remos_coverage "https://github.com/jindongwang/transferlearning/blob/master/code/deep/ReMoS/CV_adv/DNNtest/coverage/my_neuron_coverage.py"
click node_remos_wrapper "https://github.com/jindongwang/transferlearning/blob/master/code/deep/ReMoS/CV_adv/DNNtest/coverage/pytorch_wrapper.py"
click node_remos_strategy "https://github.com/jindongwang/transferlearning/blob/master/code/deep/ReMoS/CV_adv/DNNtest/strategy/random.py"
click node_remos_datasets "https://github.com/jindongwang/transferlearning/tree/master/code/deep/ReMoS/CV_adv/dataset"
click node_remos_model "https://github.com/jindongwang/transferlearning/blob/master/code/deep/ReMoS/CV_adv/model/fe_resnet.py"

classDef toneNeutral fill:#f8fafc,stroke:#334155,stroke-width:1.5px,color:#0f172a
classDef toneBlue fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
classDef toneAmber fill:#fef3c7,stroke:#d97706,stroke-width:1.5px,color:#78350f
classDef toneMint fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
classDef toneRose fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
classDef toneIndigo fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
classDef toneTeal fill:#ccfbf1,stroke:#0f766e,stroke-width:1.5px,color:#134e4a
class node_deepda_run,node_deepda_data,node_deepda_model,node_deepda_losses,node_deepda_backbones,node_daan_train,node_meda_train,node_meda_model,node_dann_train,node_dann_model toneBlue
class node_deepdg_train,node_deepdg_algorithms,node_deepdg_loaders,node_csg_runner,node_csg_methods,node_csg_arch,node_csg_data toneAmber
class node_asr_adapt,node_asr_cmatch,node_clip_runner,node_clip_model,node_clip_data,node_feature_run toneMint
class node_traditional toneRose
class node_remos_analysis,node_remos_coverage,node_remos_wrapper,node_remos_strategy,node_remos_datasets,node_remos_model toneIndigo
```

## Components

### Domain adaptation

- [DeepDA runner](https://github.com/jindongwang/transferlearning/blob/master/code/DeepDA/main.py): `code/DeepDA/main.py`
- [Domain data](https://github.com/jindongwang/transferlearning/blob/master/code/DeepDA/data_loader.py): `code/DeepDA/data_loader.py`
- [Transfer model](https://github.com/jindongwang/transferlearning/blob/master/code/DeepDA/models.py): `code/DeepDA/models.py`
- [Transfer losses](https://github.com/jindongwang/transferlearning/blob/master/code/DeepDA/transfer_losses.py): `code/DeepDA/transfer_losses.py`
- [Feature backbones](https://github.com/jindongwang/transferlearning/blob/master/code/DeepDA/backbones.py): `code/DeepDA/backbones.py`
- [DAAN training](https://github.com/jindongwang/transferlearning/blob/master/code/deep/DAAN/train.py): `code/deep/DAAN/train.py`
- [DeepMEDA training](https://github.com/jindongwang/transferlearning/blob/master/code/deep/DeepMEDA/main.py): `code/deep/DeepMEDA/main.py`
- [DeepMEDA model](https://github.com/jindongwang/transferlearning/blob/master/code/deep/DeepMEDA/deep_meda.py): `code/deep/DeepMEDA/deep_meda.py`
- [DaNN training](https://github.com/jindongwang/transferlearning/blob/master/code/deep/DaNN/main.py): `code/deep/DaNN/main.py`
- [DaNN network](https://github.com/jindongwang/transferlearning/blob/master/code/deep/DaNN/DaNN.py): `code/deep/DaNN/DaNN.py`

### Generalization methods

- [DeepDG runner](https://github.com/jindongwang/transferlearning/blob/master/code/DeepDG/train.py): `code/DeepDG/train.py`
- [DG algorithms](https://github.com/jindongwang/transferlearning/tree/master/code/DeepDG/alg/algs): `code/DeepDG/alg/algs`
- [DG data loaders](https://github.com/jindongwang/transferlearning/tree/master/code/DeepDG/datautil): `code/DeepDG/datautil`
- [CSG experiment](https://github.com/jindongwang/transferlearning/blob/master/code/deep/CSG/a-domainbed/main.py): `code/deep/CSG/a-domainbed/main.py`
- [CSG methods](https://github.com/jindongwang/transferlearning/tree/master/code/deep/CSG/methods): `code/deep/CSG/methods`
- [CSG architectures](https://github.com/jindongwang/transferlearning/blob/master/code/deep/CSG/arch/cnn.py): `code/deep/CSG/arch/cnn.py`
- [CSG data loading](https://github.com/jindongwang/transferlearning/blob/master/code/deep/CSG/utils/preprocess/data_loader.py): `code/deep/CSG/utils/preprocess/data_loader.py`

### Modalities and features

- [Speech adaptation](https://github.com/jindongwang/transferlearning/blob/master/code/ASR/Adapter/e2e_asr_adaptertransformer.py): `code/ASR/Adapter/e2e_asr_adaptertransformer.py`
- [Speech matching](https://github.com/jindongwang/transferlearning/blob/master/code/ASR/CMatch/e2e_asr_udatransformer.py): `code/ASR/CMatch/e2e_asr_udatransformer.py`
- [CLIP experiments](https://github.com/jindongwang/transferlearning/blob/master/code/clip/main.py): `code/clip/main.py`
- [CLIP model wrapper](https://github.com/jindongwang/transferlearning/blob/master/code/clip/clip_model.py): `code/clip/clip_model.py`
- [Image-text data](https://github.com/jindongwang/transferlearning/blob/master/code/clip/data/data_loader.py): `code/clip/data/data_loader.py`
- [Feature extraction](https://github.com/jindongwang/transferlearning/blob/master/code/feature_extractor/for_image_data/main.py): `code/feature_extractor/for_image_data/main.py`

### Classical methods

- [Traditional methods](https://github.com/jindongwang/transferlearning/tree/master/code/traditional): `code/traditional`

### Robustness and pruning

- [Coverage analysis](https://github.com/jindongwang/transferlearning/blob/master/code/deep/ReMoS/CV_adv/nc_prune/analyze_coverage.py): `code/deep/ReMoS/CV_adv/nc_prune/analyze_coverage.py`
- [Neuron coverage](https://github.com/jindongwang/transferlearning/blob/master/code/deep/ReMoS/CV_adv/DNNtest/coverage/my_neuron_coverage.py): `code/deep/ReMoS/CV_adv/DNNtest/coverage/my_neuron_coverage.py`
- [Model wrapper](https://github.com/jindongwang/transferlearning/blob/master/code/deep/ReMoS/CV_adv/DNNtest/coverage/pytorch_wrapper.py): `code/deep/ReMoS/CV_adv/DNNtest/coverage/pytorch_wrapper.py`
- [Neuron selection](https://github.com/jindongwang/transferlearning/blob/master/code/deep/ReMoS/CV_adv/DNNtest/strategy/random.py): `code/deep/ReMoS/CV_adv/DNNtest/strategy/random.py`
- [Vision datasets](https://github.com/jindongwang/transferlearning/tree/master/code/deep/ReMoS/CV_adv/dataset): `code/deep/ReMoS/CV_adv/dataset`
- [ResNet models](https://github.com/jindongwang/transferlearning/blob/master/code/deep/ReMoS/CV_adv/model/fe_resnet.py): `code/deep/ReMoS/CV_adv/model/fe_resnet.py`

## Connections

- DeepDA runner → Domain data: loads domains (evidence: `code/DeepDA/main.py`)
- DeepDA runner → Transfer model: builds model (evidence: `code/DeepDA/main.py`)
- Transfer model → Transfer losses: uses losses (evidence: `code/DeepDA/models.py`)
- Transfer model → Feature backbones: uses backbone (evidence: `code/DeepDA/models.py`)
- DeepDG runner → DG algorithms: selects algorithm (evidence: `code/DeepDG/train.py`)
- DeepDG runner → DG data loaders: gets loaders (evidence: `code/DeepDG/train.py`)
- CSG experiment → CSG data loading: loads data (evidence: `code/deep/CSG/a-domainbed/main.py`)
- CSG experiment → CSG methods: uses methods
- CSG experiment → CSG architectures: uses architectures
- DeepMEDA training → DeepMEDA model: builds model (evidence: `code/deep/DeepMEDA/main.py`)
- DaNN training → DaNN network: builds network (evidence: `code/deep/DaNN/main.py`)
- CLIP experiments → CLIP model wrapper: creates wrapper (evidence: `code/clip/main.py`)
- CLIP experiments → Image-text data: loads dataset (evidence: `code/clip/main.py`)
- Coverage analysis → Neuron coverage: uses coverage (evidence: `code/deep/ReMoS/CV_adv/nc_prune/analyze_coverage.py`)
- Coverage analysis → Model wrapper: wraps model (evidence: `code/deep/ReMoS/CV_adv/nc_prune/analyze_coverage.py`)
- Coverage analysis → Vision datasets: selects datasets (evidence: `code/deep/ReMoS/CV_adv/nc_prune/analyze_coverage.py`)
- Coverage analysis → ResNet models: loads network (evidence: `code/deep/ReMoS/CV_adv/nc_prune/analyze_coverage.py`)
- Coverage analysis → Neuron selection: selects neurons (evidence: `code/deep/ReMoS/CV_adv/nc_prune/analyze_coverage.py`)
