Skip to content

Commit

Permalink
contrib
Browse files Browse the repository at this point in the history
  • Loading branch information
@miguelgfierro
miguelgfierro committed Oct 28, 2019
1 parent d303314 commit fcf955f
Show file tree
Hide file tree
Showing 3 changed files with 33 additions and 29 deletions.
44 changes: 17 additions & 27 deletions README.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -47,31 +47,37 @@ To setup on your local machine:
The table below lists the recommender algorithms currently available in the repository. Notebooks are linked under the Environment column when different implementations are available.
| Algorithm | Environment | Type | Description |
| Algorithm | Environment | Type | Description |
| --- | --- | --- | --- |
| Alternating Least Squares (ALS) | [PySpark](notebooks/00_quick_start/als_movielens.ipynb) | Collaborative Filtering | Matrix factorization algorithm for explicit or implicit feedback in large datasets, optimized by Spark MLLib for scalability and distributed computing capability |
| Cornac/Bayesian Personalized Ranking (BPR) | [Python CPU](notebooks/02_model/cornac_bpr_deep_dive.ipynb) | Collaborative Filtering | Matrix factorization algorithm for predicting item ranking with implicit feedback |
| Deep Knowledge-Aware Network (DKN)<sup>*</sup> | [Python CPU / Python GPU](notebooks/00_quick_start/dkn_synthetic.ipynb) | Content-Based Filtering | Deep learning algorithm incorporating a knowledge graph and article embeddings to provide powerful news or article recommendations |
| Extreme Deep Factorization Machine (xDeepFM)<sup>*</sup> | [Python CPU / Python GPU](notebooks/00_quick_start/xdeepfm_criteo.ipynb) | Hybrid | Deep learning based algorithm for implicit and explicit feedback with user/item features |
| Alternating Least Squares (ALS) | [PySpark](notebooks/00_quick_start/als_movielens.ipynb) | Collaborative Filtering | Matrix factorization algorithm for explicit or implicit feedback in large datasets, optimized by Spark MLLib for scalability and distributed computing capability |
| Cornac/Bayesian Personalized Ranking (BPR) | [Python CPU](notebooks/02_model/cornac_bpr_deep_dive.ipynb) | Collaborative Filtering | Matrix factorization algorithm for predicting item ranking with implicit feedback |
| Deep Knowledge-Aware Network (DKN)<sup>*</sup> | [Python CPU / Python GPU](notebooks/00_quick_start/dkn_synthetic.ipynb) | Content-Based Filtering | Deep learning algorithm incorporating a knowledge graph and article embeddings to provide powerful news or article recommendations |
| Extreme Deep Factorization Machine (xDeepFM)<sup>*</sup> | [Python CPU / Python GPU](notebooks/00_quick_start/xdeepfm_criteo.ipynb) | Hybrid | Deep learning based algorithm for implicit and explicit feedback with user/item features |
| FastAI Embedding Dot Bias (FAST) | [Python CPU / Python GPU](notebooks/00_quick_start/fastai_movielens.ipynb) | Collaborative Filtering | General purpose algorithm with embeddings and biases for users and items |
| LightGBM/Gradient Boosting Tree<sup>*</sup> | [Python CPU](notebooks/00_quick_start/lightgbm_tinycriteo.ipynb) / [PySpark](notebooks/02_model/mmlspark_lightgbm_criteo.ipynb) | Content-Based Filtering | Gradient Boosting Tree algorithm for fast training and low memory usage in content-based problems |
| Neural Collaborative Filtering (NCF) | [Python CPU / Python GPU](notebooks/00_quick_start/ncf_movielens.ipynb) | Collaborative Filtering | Deep learning algorithm with enhanced performance for implicit feedback |
| Restricted Boltzmann Machines (RBM) | [Python CPU / Python GPU](notebooks/00_quick_start/rbm_movielens.ipynb) | Collaborative Filtering | Neural network based algorithm for learning the underlying probability distribution for explicit or implicit feedback |
| Neural Collaborative Filtering (NCF) | [Python CPU / Python GPU](notebooks/00_quick_start/ncf_movielens.ipynb) | Collaborative Filtering | Deep learning algorithm with enhanced performance for implicit feedback |
| Restricted Boltzmann Machines (RBM) | [Python CPU / Python GPU](notebooks/00_quick_start/rbm_movielens.ipynb) | Collaborative Filtering | Neural network based algorithm for learning the underlying probability distribution for explicit or implicit feedback |
| Riemannian Low-rank Matrix Completion (RLRMC)<sup>*</sup> | [Python CPU](notebooks/00_quick_start/rlrmc_movielens.ipynb) | Collaborative Filtering | Matrix factorization algorithm using Riemannian conjugate gradients optimization with small memory consumption. |
| Simple Algorithm for Recommendation (SAR)<sup>*</sup> | [Python CPU](notebooks/00_quick_start/sar_movielens.ipynb) | Collaborative Filtering | Similarity-based algorithm for implicit feedback dataset |
| Surprise/Singular Value Decomposition (SVD) | [Python CPU](notebooks/02_model/surprise_svd_deep_dive.ipynb) | Collaborative Filtering | Matrix factorization algorithm for predicting explicit rating feedback in datasets that are not very large |
| Surprise/Singular Value Decomposition (SVD) | [Python CPU](notebooks/02_model/surprise_svd_deep_dive.ipynb) | Collaborative Filtering | Matrix factorization algorithm for predicting explicit rating feedback in datasets that are not very large |
| Vowpal Wabbit Family (VW)<sup>*</sup> | [Python CPU (online training)](notebooks/02_model/vowpal_wabbit_deep_dive.ipynb) | Content-Based Filtering | Fast online learning algorithms, great for scenarios where user features / context are constantly changing |
| Wide and Deep | [Python CPU / Python GPU](notebooks/00_quick_start/wide_deep_movielens.ipynb) | Hybrid | Deep learning algorithm that can memorize feature interactions and generalize user features |
**NOTE**: <sup>*</sup> indicates algorithms invented/contributed by Microsoft.
**Preliminary Comparison**
Independent or incubating algorithms and utilities are candidates for the [contrib](contrib) folder. This will house contributions which may not easily fit into the core repository or need time to refactor or mature the code and add necessary tests.
| Algorithm | Environment | Type | Description |
| --- | --- | --- | --- |
| [SARplus](contrib/sarplus/README.md) | PySpark | Collaborative Filtering | Optimized implementation of SAR for Spark |
### Preliminary Comparison
We provide a [benchmark notebook](benchmarks/movielens.ipynb) to illustrate how different algorithms could be evaluated and compared. In this notebook, the MovieLens dataset is split into training/test sets at a 75/25 ratio using a stratified split. A recommendation model is trained using each of the collaborative filtering algorithms below. We utilize empirical parameter values reported in literature [here](http://mymedialite.net/examples/datasets.html). For ranking metrics we use `k=10` (top 10 recommended items). We run the comparison on a Standard NC6s_v2 [Azure DSVM](https://azure.microsoft.com/en-us/services/virtual-machines/data-science-virtual-machines/) (6 vCPUs, 112 GB memory and 1 P100 GPU). Spark ALS is run in local standalone mode. In this table we show the results on Movielens 100k, running the algorithms for 15 epochs.
| Algo | MAP | nDCG@k | Precision@k | Recall@k | RMSE | MAE | R<sup>2</sup> | Explained Variance |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| [ALS](notebooks/00_quick_start/als_movielens.ipynb) | 0.004732 | 0.044239 | 0.048462 | 0.017796 | 0.965038 | 0.753001 | 0.255647 | 0.251648 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| [ALS](notebooks/00_quick_start/als_movielens.ipynb) | 0.004732 | 0.044239 | 0.048462 | 0.017796 | 0.965038 | 0.753001 | 0.255647 | 0.251648 |
| [SVD](notebooks/02_model/surprise_svd_deep_dive.ipynb) | 0.012873 | 0.095930 | 0.091198 | 0.032783 | 0.938681 | 0.742690 | 0.291967 | 0.291971 |
| [SAR](notebooks/00_quick_start/sar_movielens.ipynb) | 0.113028 | 0.388321 | 0.333828 | 0.183179 | N/A | N/A | N/A | N/A |
| [NCF](notebooks/02_model/ncf_deep_dive.ipynb) | 0.107720 | 0.396118 | 0.347296 | 0.180775 | N/A | N/A | N/A | N/A |
Expand Down Expand Up @@ -107,19 +113,3 @@ The following tests run on an AzureML [compute target](https://docs.microsoft.co
| --- | --- | --- | --- | --- | --- |
| **nightly_cpu_tests** | master | [![Build Status](https://dev.azure.com/best-practices/recommenders/_apis/build/status/nightly_cpu_tests?branchName=master)](https://dev.azure.com/best-practices/recommenders/_build/latest?definitionId=25&branchName=master) | | Staging | [![Build Status](https://dev.azure.com/best-practices/recommenders/_apis/build/status/nightly_cpu_tests?branchName=staging)](https://dev.azure.com/best-practices/recommenders/_build/latest?definitionId=25&branchName=staging) |
| **nightly_gpu_tests** | master | [![Build Status](https://dev.azure.com/best-practices/recommenders/_apis/build/status/bp-nightly_gpu_tests?branchName=master)](https://dev.azure.com/best-practices/recommenders/_build/latest?definitionId=5&branchName=master) | | Staging | [![Build Status](https://dev.azure.com/best-practices/recommenders/_apis/build/status/bp-nightly_gpu_tests?branchName=staging)](https://dev.azure.com/best-practices/recommenders/_build/latest?definitionId=5&branchName=staging) |
### Contrib
Independent or incubating algorithms and utilities are candidates for the [contrib](contrib) folder. This will house contributions which may not easily fit into the core repository or need time to refactor or mature the code and add necessary tests.
| Algorithm | Environment | Type | Description |
| --- | --- | --- | --- |
| [SARplus](contrib/sarplus/README.md) | PySpark | Collaborative Filtering | Optimized implementation of SAR for Spark |
### Contrib Build status
| Build Type | Branch | Status |
| --- | --- | --- |
| **nightly** | master | [![Build Status](https://dev.azure.com/best-practices/recommenders/_apis/build/status/contrib%20sarplus?branchName=master)](https://dev.azure.com/best-practices/recommenders/_build/latest?definitionId=107&branchName=master) |
13 changes: 13 additions & 0 deletions contrib/README.md
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,13 @@
# Contrib

Independent or incubating algorithms and utilities are candidates for the `contrib` folder. This will house contributions which may not easily fit into the core repository or need time to refactor or mature the code and add necessary tests.

| Name | Description |
|------|-------------|
| [SARplus](sarplus/README.md) | Optimized implementation of SAR for Spark |

## Contrib Build status

| Name | Status |
|------|--------|
| **SARplus** | [![Build Status](https://dev.azure.com/best-practices/recommenders/_apis/build/status/contrib%20sarplus?branchName=master)](https://dev.azure.com/best-practices/recommenders/_build/latest?definitionId=107&branchName=master) |
5 changes: 3 additions & 2 deletions contrib/sarplus/README.md
View file
Original file line number Diff line number Diff line change
@@ -1,5 +1,6 @@
# SARplus
pronounced sUrplus as it's simply better if not best!

Pronounced surplus as it's simply better if not best!

[![Build Status](https://dev.azure.com/best-practices/recommenders/_apis/build/status/contrib%20sarplus?branchName=master)](https://dev.azure.com/best-practices/recommenders/_build/latest?definitionId=107&branchName=master)
[![PyPI version](https://badge.fury.io/py/pysarplus.svg)](https://badge.fury.io/py/pysarplus)
Expand Down Expand Up @@ -29,7 +30,7 @@ There are a couple of key optimizations:
** make sure to always just keep top-k items in-memory
** use standard join using binary search between users past seen items and the related items

![Image of sarplus top-k recommendation optimization](https://recodatasets.blob.core.windows.net/images/sarplus_udf.svg)
![Image of sarplus top-k recommendation optimization](https://recodatasets.blob.core.windows.net/images/sarplus_udf.svg)

# Usage

Expand Down

0 comments on commit fcf955f

Please sign in to comment.