This is the MATLAB code for deep Gaussian process for inverse reinforcement learning (DGP-IRL) algorithm. DGP-IRL can efficiently infer about agent intention by observing its actions in the world, and learn to act in a similar way as the agent.
Inverse reinforcement learning via deep Gaussian Process
UAI (2017) (paper | code | supplementary)
Sensing by proxy estimates the number of people in the room using CO2 sensors. It's more accurate than previous machine learning models, and could be used to improve the efficiency of demand-controlled ventilation systems.
Occupancy detection via environmental sensing
IEEE Transactions on Automation Science and Engineering (TASE) (2017) (paper | code and data)
See also "Sensing by proxy: Occupancy detection based on indoor CO2 concentration" (Best Paper Award)
Featured in CO2Meter.com article (Feb 2017) CO2 Sensor Occupancy Detection
Microgrid Optimization with District Energy Systems Tool (MODEST) is Python-based software to optimize the *planning and operation of a microgrid* to improve economics and resilience. It also *optimizes the energy retail rates in a district* to enhance retailer profit and system reliability.
Please contact [Ming Jin](mailto:firstname.lastname@example.org) and [Wei Feng](mailto:WeiFeng@lbl.gov) for further details.
See papers: Microgrid to enable optimal distributed energy retail and end-user demand response, Applied Energy (2017) (paper)
MOD-DR: Microgrid optimal dispatch with demand response, Applied Energy, 187, 758 - 776 (2017)(paper | poster)
Experiment can be expensive. That's why we need to determine the duration of the experiment ahead of time. This is MATLAB implementation of the reliable estimation of stopping time (REST) algorithm. Given the precision and confidence requirements, it produces a statistically guaranteed stopping time, so it is easier to manage resources and project budget.
REST: Reliable estimation of stopping time algorithm for social game experiments
ACM/IEEE ICCPS (2015) (paper | code)
This is Python implementation of spatio-temporal interpolation methods used to process mobile sensing data to generate a continuous 3D evolutionary map of the *indoor environmental quality*. Also includes software implementation of robot operation system (ROS) code for TurtleBot2, and Arduino-based environmental sensing platform.
Automated mobile sensing: Towards high-granularity agile indoor environmental quality monitoring
Building and Environment (2017) (paper | slide | code)
IPython for [EE 120: Signals ans Systems](https://inst.eecs.berkeley.edu/~ee120/fa16/), that employs Fourier Transform, bandpass filter, and up/down sampling to magnify subtle changes in a video to [reveal pulses and baby breathing](https://www.youtube.com/watch?v=e9ASH8IBJ2U). Designed to apply basic toolset to solve real problems.
Here are some [concept questions](/archive/python_evm.pdf) for teaching.
See the original paper "Eulerian Video Magnification for Revealing Subtle Changes in the World"