HYPPO is a software package that uses adaptive surrogate models with uncertainty quantification to automatically tune hyperparameters in a variety of deep learning models. The hyperparamer optimization (HPO) process uses the following steps:

Initialization : Maps architecture to performance of trained model
Prediction : Choose which architecture to try next using surrogate modeling
Determine model’s accuracy using test data
Iteratively update with new input-output data

This HPO problem is formulated as a bilevel optimization problem:

\[\begin{split}\min_{\theta,w^*}u(\theta,w^*;D_{val}) \,\, \text{s.t.}\,\, \theta \in \Omega \,\,\,\, (1)\\ w* \in \text{arg}\min_{w\in W}\mathcal{l}(w;\theta,D_{train}) \,\,\,\, (2)\end{split}\]

The upper level problem (1) finds the optimal set of hyperparameters (i.e., the model architecture) that minimizes the objective function. The lower level problem (2) finds the optimal weights that minimze the loss function for the training data.

This documentation aims at providing a comprehensive background and instruction for the HYPPO software. The program can be used for a variety of applications, including but not limited to performing Deep Learning training across many sets of hyperparameters with multiple trials for each evaluation, doing sensitivity analysis on sparse data across the hyperparameter space, and doing surrogate modeling for hyperparameter optimization.

Installation ¶

The HYPPO software can be installed very easily using the pip Python package manager as follows:

          pip install hpo-uq

         

The above command will install all the required dependencies (see list below) along with the main software. More information about the package can be found in the pypi webpage .

          deap>=1.3
matplotlib
numpy>=1.19.2
pandas>=1.1.3
pickle5>=0.0.11
plotly>=4.14.3
pyyaml>=5.3.1
SALib>=1.3.13
scipy
sklearn

         

HPO Solutions ¶

Below is an example of how HPO model solutions provided by HYPPO can be visualized. The figure below shows the solutions obtained by applying HYPPO to a time-series prediction problem. Please refer to the HYPPO introductory paper for more details. To reproduce the results from this paper, please visit this Google Colab notebook .

License Agreement ¶

          *** License Agreement ***

HYPPO Copyright (c) 2021, The Regents of the University of California,
through Lawrence Berkeley National Laboratory (subject to receipt of
any required approvals from the U.S. Dept. of Energy). All rights reserved.

Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:

(1) Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.

(2) Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.

(3) Neither the name of the University of California, Lawrence Berkeley
National Laboratory, U.S. Dept. of Energy nor the names of its contributors
may be used to endorse or promote products derived from this software
without specific prior written permission.


THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.

You are under no obligation whatsoever to provide any bug fixes, patches,
or upgrades to the features, functionality or performance of the source
code ("Enhancements") to anyone; however, if you choose to make your
Enhancements available either publicly, or directly to Lawrence Berkeley
National Laboratory, without imposing a separate written license agreement
for such Enhancements, then you hereby grant the following license: a
non-exclusive, royalty-free perpetual license to install, use, modify,
prepare derivative works, incorporate into other computer software,
distribute, and sublicense such enhancements or derivative works thereof,
in binary and source code form.

         

Reporting issues ¶

Submit a ticket

If you find any bugs when running this program, please make sure to report them by clicking on the above link and submit a ticket on the software’s official Gitlab repository.