# photodetectors <br> MUTC Photodetector Dataset <br> This dataset lists four performance metrics of 1755 MUTC photodetectors with 17 layers. <br> The semiconductor types of these layers are fixed and they can be found in https://doi.org/10.1364/OE.27.003717 <br> **Columns** phasenoise : true phasenoise of the photodetector (Unit: dBc/Hz) <br> phasenoise_15mA : hypothetical phasenoise of the photodetector if the output current was equal to 15 mA (Unit: dBc/Hz) <br> current : average current of the photodetector (Unit: mA) <br> IR_max : maximum value of the impulse response of the photodetector <br> decay_time: the time it takes for the impulse response to decay from its maximum value to 1% of that maximum value (Unit: ns) <br> t1 ... t17 : thickness each layer (Unit: nm) <br> d1 ... d17 : doping concentration of each layer (Unit: cm^-3) <br> **Forward Problem:** Predicting Performance Metric From Design Parameters <br> The four machine learning algorithms, linear regression, k-nearest neighbor, random forest, and an artificial neural network, are used to predict performance metrics (phasenoise, current, IR_Max, and decay_time) from the design parameters (thicknesses and doping concentrations)