Fix MTBO one-hot encoding bug + improve docs (#116)

docs/source/strategies.rst

@@ -21,14 +21,15 @@ With the updated model, we use one of two classes of techniques to select our ne
    :height: 300 px
    :figclass: align-center
 
-   Illustration of how acquisition functions eanble BO strategies to reduce uncertainty and maximise objective simulataneously. Dotted line is actual objective and solid line is posterior of surrogate model. Acquisition function is high where objective to be optimal (exploration) and where there is high uncertainty (exploitation). Adapted from `Shahriari et al. <https://www.cs.ox.ac.uk/people/nando.defreitas/publications/BayesOptLoop.pdf>`_
+   Illustration of how acquisition functions eanble BO strategies to reduce uncertainty and maximise objective simulataneously. Dotted line is actual objective and solid line is posterior of surrogate model. Acquisition function is high where objective to be optimal (exploration) and where there is high uncertainty (exploitation). Adapted from `Shahriari et al. <https://ieeexplore.ieee.org/document/7352306>`_
 
-To learn more about BO, we suggest reading the review by `Shahriari et al. <https://www.cs.ox.ac.uk/people/nando.defreitas/publications/BayesOptLoop.pdf>`_
+To learn more about BO, we suggest reading the review by `Shahriari et al. <https://ieeexplore.ieee.org/document/7352306>`_
 
 The BO strategies available in Summit are:
 
 * TSEMO_
 * SOBO_
+* MTBO_
 * ENTMOOT_
 
 
@@ -49,6 +50,16 @@ SOBO
 .. autoclass:: summit.strategies.sobo.SOBO
    :members:
 
+
+MTBO
+^^^^
+
+.. _MTBO:
+
+.. autoclass:: summit.strategies.MTBO
+    :members:
+    :inherited-members:
+
 ENTMOOT
 ^^^^^^^
 
@@ -57,6 +68,7 @@ ENTMOOT
 .. autoclass:: summit.strategies.ENTMOOT
    :members:
 
+
 Reinforcement Learning
 ----------------------
 

experiments/multitask/baumgartner_phenethylamine_emulator/baumgartner_primary_amine.json

@@ -0,0 +1 @@
+{"domain": [{"type": "CategoricalVariable", "is_objective": false, "name": "catalyst", "description": "Catalyst type", "units": null, "levels": ["tBuXPhos", "tBuBrettPhos", "AlPhos"], "ds": {"index": ["tBuXPhos", "tBuBrettPhos", "AlPhos"], "columns": [["area_cat", "DATA"], ["M2_cat", "DATA"]], "data": [[460.7543, 67.2057], [518.8408, 89.8738], [819.933, 129.0808]]}}, {"type": "CategoricalVariable", "is_objective": false, "name": "base", "description": "Base", "units": null, "levels": ["BTTP", "BTMG", "DBU", "MTBD"], "ds": null}, {"type": "ContinuousVariable", "is_objective": false, "name": "base_equivalents", "description": "Base equivalents", "units": null, "bounds": [1.0, 2.5]}, {"type": "ContinuousVariable", "is_objective": false, "name": "temperature", "description": "Temperature in degrees Celsius (\u00baC)", "units": null, "bounds": [30.0, 100.0]}, {"type": "ContinuousVariable", "is_objective": false, "name": "t_res", "description": "residence time in seconds (s)", "units": null, "bounds": [60.0, 1800.0]}, {"type": "ContinuousVariable", "is_objective": true, "name": "yield", "description": "Yield", "units": null, "bounds": [0.0, 1.0]}], "name": "ExperimentalEmulator", "data": {"index": [], "columns": [["catalyst", "DATA"], ["base", "DATA"], ["base_equivalents", "DATA"], ["temperature", "DATA"], ["t_res", "DATA"], ["yield", "DATA"], ["computation_t", "METADATA"], ["experiment_t", "METADATA"], ["strategy", "METADATA"]], "data": []}, "experiment_params": {"model_name": "baumgartner_primary_amine", "regressor_name": "ANNRegressor", "n_features": 10, "n_examples": 92, "descriptors_features": [], "output_variable_names": ["yield"], "predictors": [{"input_preprocessor": {"num": {"mean_": [1.6433899101513958, 77.7690909090909, 732.7608206142079], "var_": [0.12978876347780144, 753.5497719008265, 371643.348125646], "scale_": [0.3602620761026637, 27.45086104115546, 609.6255802750128], "n_samples_seen_": 55}}, "output_preprocessor": {"mean_": [0.4522085499145429], "var_": [0.15275946137050844], "scale_": [0.39084454885607456], "n_samples_seen_": 55}}, {"input_preprocessor": {"num": {"mean_": [1.6189573775076675, 74.81272727272727, 810.9457470980558], "var_": [0.13795978110802035, 832.2982016528924, 384193.0280846722], "scale_": [0.37142937566651935, 28.849578881725336, 619.8330646913507], "n_samples_seen_": 55}}, "output_preprocessor": {"mean_": [0.48744336173404007], "var_": [0.14849834688886238], "scale_": [0.3853548324451925], "n_samples_seen_": 55}}, {"input_preprocessor": {"num": {"mean_": [1.6285175488857782, 79.79818181818183, 781.0131804813038], "var_": [0.13940704012600072, 693.8216330578513, 389947.7201171372], "scale_": [0.37337252192147286, 26.340494168824, 624.4579410313694], "n_samples_seen_": 55}}, "output_preprocessor": {"mean_": [0.4867995243329047], "var_": [0.1458822295239026], "scale_": [0.38194532268886683], "n_samples_seen_": 55}}, {"input_preprocessor": {"num": {"mean_": [1.5791994991469978, 76.1381818181818, 800.1719308419661], "var_": [0.14730072854572404, 761.1507239669419, 394083.7678590847], "scale_": [0.3837977703761762, 27.588960182778578, 627.7609161608301], "n_samples_seen_": 55}}, "output_preprocessor": {"mean_": [0.4951012723190202], "var_": [0.1582074359916574], "scale_": [0.3977529836363989], "n_samples_seen_": 55}}, {"input_preprocessor": {"num": {"mean_": [1.6307636010361506, 75.75535714285715, 832.615480014256], "var_": [0.13706793334467066, 768.2367570153064, 385465.5730380972], "scale_": [0.3702268673998021, 27.717084208395846, 620.8587383923151], "n_samples_seen_": 56}}, "output_preprocessor": {"mean_": [0.44402859465169187], "var_": [0.16073403177350493], "scale_": [0.40091648977499655], "n_samples_seen_": 56}}]}, "extras": []}

experiments/multitask/data/cross_coupling_similar/mtbo_pre-train_repeat_0.json

@@ -0,0 +1 @@
+{"runner": {"num_initial_experiments": null, "max_iterations": 20, "batch_size": 1, "f_tol": 1e-05, "max_restarts": 0}, "strategy": {"name": "MTBO", "transform": {"transform_domain": [{"type": "CategoricalVariable", "is_objective": false, "name": "catalyst", "description": "Catalyst type", "units": null, "levels": ["tBuXPhos", "tBuBrettPhos", "AlPhos"], "ds": {"index": ["tBuXPhos", "tBuBrettPhos", "AlPhos"], "columns": [["area_cat", "DATA"], ["M2_cat", "DATA"]], "data": [[460.7543, 67.2057], [518.8408, 89.8738], [819.933, 129.0808]]}}, {"type": "CategoricalVariable", "is_objective": false, "name": "base", "description": "Base", "units": null, "levels": ["BTTP", "BTMG", "DBU", "MTBD"], "ds": null}, {"type": "ContinuousVariable", "is_objective": false, "name": "base_equivalents", "description": "Base equivalents", "units": null, "bounds": [1.0, 2.5]}, {"type": "ContinuousVariable", "is_objective": false, "name": "temperature", "description": "Temperature in degrees Celsius (\u00baC)", "units": null, "bounds": [30.0, 100.0]}, {"type": "ContinuousVariable", "is_objective": false, "name": "t_res", "description": "residence time in seconds (s)", "units": null, "bounds": [60.0, 1800.0]}, {"type": "ContinuousVariable", "is_objective": true, "name": "yield", "description": "Yield", "units": null, "bounds": [0.0, 1.0]}], "name": "Transform", "domain": [{"type": "CategoricalVariable", "is_objective": false, "name": "catalyst", "description": "Catalyst type", "units": null, "levels": ["tBuXPhos", "tBuBrettPhos", "AlPhos"], "ds": {"index": ["tBuXPhos", "tBuBrettPhos", "AlPhos"], "columns": [["area_cat", "DATA"], ["M2_cat", "DATA"]], "data": [[460.7543, 67.2057], [518.8408, 89.8738], [819.933, 129.0808]]}}, {"type": "CategoricalVariable", "is_objective": false, "name": "base", "description": "Base", "units": null, "levels": ["BTTP", "BTMG", "DBU", "MTBD"], "ds": null}, {"type": "ContinuousVariable", "is_objective": false, "name": "base_equivalents", "description": "Base equivalents", "units": null, "bounds": [1.0, 2.5]}, {"type": "ContinuousVariable", "is_objective": false, "name": "temperature", "description": "Temperature in degrees Celsius (\u00baC)", "units": null, "bounds": [30.0, 100.0]}, {"type": "ContinuousVariable", "is_objective": false, "name": "t_res", "description": "residence time in seconds (s)", "units": null, "bounds": [60.0, 1800.0]}, {"type": "ContinuousVariable", "is_objective": true, "name": "yield", "description": "Yield", "units": null, "bounds": [0.0, 1.0]}], "transform_params": {}}, "strategy_params": {"all_experiments": {"index": [0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], "columns": [["catalyst", "DATA"], ["base", "DATA"], ["base_equivalents", "DATA"], ["temperature", "DATA"], ["t_res", "DATA"], ["yield", "DATA"], ["computation_t", "METADATA"], ["experiment_t", "METADATA"], ["strategy", "METADATA"], ["task", "METADATA"]], "data": [["tBuXPhos", "BTTP", 2.125, 82.5, 495.0, 0.8031633496284485, 0.0, 0.039190053939819336, "LHS", 1.0], ["AlPhos", "BTTP", 1.375, 47.5, 1365.0, 0.932254433631897, 0.0, 0.03976082801818848, "LHS", 1.0], ["AlPhos", "BTTP", 1.163477897644043, 68.20264434814453, 1800.0, 0.9977060556411743, 3.4609227180480957, 0.030591964721679688, "MTBO", 1.0], ["AlPhos", "BTTP", 1.1810085773468018, 65.93614196777344, 1800.0, 0.9983388781547546, 1.8297340869903564, 0.038281917572021484, "MTBO", 1.0], ["AlPhos", "BTTP", 1.200371265411377, 67.4896240234375, 1800.0, 1.0, 2.0852811336517334, 0.030476093292236328, "MTBO", 1.0], ["AlPhos", "BTTP", 1.431157112121582, 68.3565902709961, 1671.76025390625, 1.0, 1.7265970706939697, 0.030106067657470703, "MTBO", 1.0], ["AlPhos", "BTTP", 1.2855861186981201, 64.85208129882812, 1745.18212890625, 1.0, 3.839944839477539, 0.038710594177246094, "MTBO", 1.0], ["AlPhos", "BTTP", 1.3503124713897705, 67.24772644042969, 1719.7197265625, 1.0, 2.4471189975738525, 0.03168916702270508, "MTBO", 1.0], ["AlPhos", "BTTP", 1.1549394130706787, 66.33074951171875, 1800.0, 0.9959722757339478, 2.9563791751861572, 0.028990983963012695, "MTBO", 1.0], ["AlPhos", "BTTP", 1.33616042137146, 68.44651794433594, 1797.897216796875, 1.0, 4.110234260559082, 0.03172802925109863, "MTBO", 1.0], ["AlPhos", "BTTP", 1.5323737859725952, 68.9630126953125, 1693.62109375, 1.0, 2.1891300678253174, 0.03180408477783203, "MTBO", 1.0], ["AlPhos", "BTTP", 1.3426467180252075, 66.57575988769531, 1738.040283203125, 1.0, 1.6926929950714111, 0.030109882354736328, "MTBO", 1.0], ["AlPhos", "BTTP", 1.3669359683990479, 66.5360107421875, 1395.382568359375, 1.0, 2.829677104949951, 0.03584575653076172, "MTBO", 1.0], ["AlPhos", "BTTP", 1.5615137815475464, 53.81058883666992, 1359.92578125, 0.9918607473373413, 4.797755002975464, 0.045375823974609375, "MTBO", 1.0], ["AlPhos", "BTTP", 1.5345615148544312, 61.66748046875, 1549.867919921875, 1.0, 4.9743812084198, 0.03585195541381836, "MTBO", 1.0], ["AlPhos", "BTTP", 1.2645065784454346, 61.865821838378906, 1752.38134765625, 1.0, 2.0327069759368896, 0.03057098388671875, "MTBO", 1.0], ["AlPhos", "BTTP", 1.271968960762024, 67.43891906738281, 1262.8048095703125, 0.9663092494010925, 2.946610927581787, 0.04000687599182129, "MTBO", 1.0], ["AlPhos", "BTTP", 1.2886420488357544, 63.50016403198242, 1514.847412109375, 0.9973722696304321, 2.6807632446289062, 0.030769824981689453, "MTBO", 1.0], ["AlPhos", "BTTP", 1.4443421363830566, 59.67304229736328, 1362.010498046875, 0.9968655705451965, 3.1751291751861572, 0.03396105766296387, "MTBO", 1.0], ["AlPhos", "BTTP", 1.4641574621200562, 52.60327911376953, 1514.437744140625, 1.0, 1.9494881629943848, 0.02869415283203125, "MTBO", 1.0]]}, "categorical_method": "one-hot", "task": 1}}, "experiment": {"domain": [{"type": "CategoricalVariable", "is_objective": false, "name": "catalyst", "description": "Catalyst type", "units": null, "levels": ["tBuXPhos", "tBuBrettPhos", "AlPhos"], "ds": {"index": ["tBuXPhos", "tBuBrettPhos", "AlPhos"], "columns": [["area_cat", "DATA"], ["M2_cat", "DATA"]], "data": [[460.7543, 67.2057], [518.8408, 89.8738], [819.933, 129.0808]]}}, {"type": "CategoricalVariable", "is_objective": false, "name": "base", "description": "Base", "units": null, "levels": ["BTTP", "BTMG", "DBU", "MTBD"], "ds": null}, {"type": "ContinuousVariable", "is_objective": false, "name": "base_equivalents", "description": "Base equivalents", "units": null, "bounds": [1.0, 2.5]}, {"type": "ContinuousVariable", "is_objective": false, "name": "temperature", "description": "Temperature in degrees Celsius (\u00baC)", "units": null, "bounds": [30.0, 100.0]}, {"type": "ContinuousVariable", "is_objective": false, "name": "t_res", "description": "residence time in seconds (s)", "units": null, "bounds": [60.0, 1800.0]}, {"type": "ContinuousVariable", "is_objective": true, "name": "yield", "description": "Yield", "units": null, "bounds": [0.0, 1.0]}], "name": "ExperimentalEmulator", "data": {"index": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20], "columns": [["catalyst", "DATA"], ["base", "DATA"], ["base_equivalents", "DATA"], ["temperature", "DATA"], ["t_res", "DATA"], ["yield", "DATA"], ["computation_t", "METADATA"], ["experiment_t", "METADATA"], ["strategy", "METADATA"], ["task", "METADATA"]], "data": [["tBuXPhos", "BTTP", 2.125, 82.5, 495.0, 0.8031633496284485, 0.0, 0.039190053939819336, "LHS", 1.0], ["AlPhos", "BTTP", 1.375, 47.5, 1365.0, 0.932254433631897, 0.0, 0.03976082801818848, "LHS", 1.0], ["AlPhos", "BTTP", 1.163477897644043, 68.20264434814453, 1800.0, 0.9977060556411743, 3.4609227180480957, 0.030591964721679688, "MTBO", 1.0], ["AlPhos", "BTTP", 1.1810085773468018, 65.93614196777344, 1800.0, 0.9983388781547546, 1.8297340869903564, 0.038281917572021484, "MTBO", 1.0], ["AlPhos", "BTTP", 1.200371265411377, 67.4896240234375, 1800.0, 1.0, 2.0852811336517334, 0.030476093292236328, "MTBO", 1.0], ["AlPhos", "BTTP", 1.431157112121582, 68.3565902709961, 1671.76025390625, 1.0, 1.7265970706939697, 0.030106067657470703, "MTBO", 1.0], ["AlPhos", "BTTP", 1.2855861186981201, 64.85208129882812, 1745.18212890625, 1.0, 3.839944839477539, 0.038710594177246094, "MTBO", 1.0], ["AlPhos", "BTTP", 1.3503124713897705, 67.24772644042969, 1719.7197265625, 1.0, 2.4471189975738525, 0.03168916702270508, "MTBO", 1.0], ["AlPhos", "BTTP", 1.1549394130706787, 66.33074951171875, 1800.0, 0.9959722757339478, 2.9563791751861572, 0.028990983963012695, "MTBO", 1.0], ["AlPhos", "BTTP", 1.33616042137146, 68.44651794433594, 1797.897216796875, 1.0, 4.110234260559082, 0.03172802925109863, "MTBO", 1.0], ["AlPhos", "BTTP", 1.5323737859725952, 68.9630126953125, 1693.62109375, 1.0, 2.1891300678253174, 0.03180408477783203, "MTBO", 1.0], ["AlPhos", "BTTP", 1.3426467180252075, 66.57575988769531, 1738.040283203125, 1.0, 1.6926929950714111, 0.030109882354736328, "MTBO", 1.0], ["AlPhos", "BTTP", 1.3669359683990479, 66.5360107421875, 1395.382568359375, 1.0, 2.829677104949951, 0.03584575653076172, "MTBO", 1.0], ["AlPhos", "BTTP", 1.5615137815475464, 53.81058883666992, 1359.92578125, 0.9918607473373413, 4.797755002975464, 0.045375823974609375, "MTBO", 1.0], ["AlPhos", "BTTP", 1.5345615148544312, 61.66748046875, 1549.867919921875, 1.0, 4.9743812084198, 0.03585195541381836, "MTBO", 1.0], ["AlPhos", "BTTP", 1.2645065784454346, 61.865821838378906, 1752.38134765625, 1.0, 2.0327069759368896, 0.03057098388671875, "MTBO", 1.0], ["AlPhos", "BTTP", 1.271968960762024, 67.43891906738281, 1262.8048095703125, 0.9663092494010925, 2.946610927581787, 0.04000687599182129, "MTBO", 1.0], ["AlPhos", "BTTP", 1.2886420488357544, 63.50016403198242, 1514.847412109375, 0.9973722696304321, 2.6807632446289062, 0.030769824981689453, "MTBO", 1.0], ["AlPhos", "BTTP", 1.4443421363830566, 59.67304229736328, 1362.010498046875, 0.9968655705451965, 3.1751291751861572, 0.03396105766296387, "MTBO", 1.0], ["AlPhos", "BTTP", 1.4641574621200562, 52.60327911376953, 1514.437744140625, 1.0, 1.9494881629943848, 0.02869415283203125, "MTBO", 1.0], ["AlPhos", "BTTP", 1.463046908378601, 57.9471549987793, 1606.17919921875, 1.0, 3.123176097869873, 0.030063867568969727, "MTBO", 1.0]]}, "experiment_params": {"model_name": "baumgartner_primary_amine", "regressor_name": "ANNRegressor", "n_features": 10, "n_examples": 92, "descriptors_features": [], "output_variable_names": ["yield"], "predictors": [{"input_preprocessor": {"num": {"mean_": [1.6433899101513958, 77.7690909090909, 732.7608206142079], "var_": [0.12978876347780144, 753.5497719008265, 371643.348125646], "scale_": [0.3602620761026637, 27.45086104115546, 609.6255802750128], "n_samples_seen_": 55}}, "output_preprocessor": {"mean_": [0.4522085499145429], "var_": [0.15275946137050844], "scale_": [0.39084454885607456], "n_samples_seen_": 55}}, {"input_preprocessor": {"num": {"mean_": [1.6189573775076675, 74.81272727272727, 810.9457470980558], "var_": [0.13795978110802035, 832.2982016528924, 384193.0280846722], "scale_": [0.37142937566651935, 28.849578881725336, 619.8330646913507], "n_samples_seen_": 55}}, "output_preprocessor": {"mean_": [0.48744336173404007], "var_": [0.14849834688886238], "scale_": [0.3853548324451925], "n_samples_seen_": 55}}, {"input_preprocessor": {"num": {"mean_": [1.6285175488857782, 79.79818181818183, 781.0131804813038], "var_": [0.13940704012600072, 693.8216330578513, 389947.7201171372], "scale_": [0.37337252192147286, 26.340494168824, 624.4579410313694], "n_samples_seen_": 55}}, "output_preprocessor": {"mean_": [0.4867995243329047], "var_": [0.1458822295239026], "scale_": [0.38194532268886683], "n_samples_seen_": 55}}, {"input_preprocessor": {"num": {"mean_": [1.5791994991469978, 76.1381818181818, 800.1719308419661], "var_": [0.14730072854572404, 761.1507239669419, 394083.7678590847], "scale_": [0.3837977703761762, 27.588960182778578, 627.7609161608301], "n_samples_seen_": 55}}, "output_preprocessor": {"mean_": [0.4951012723190202], "var_": [0.1582074359916574], "scale_": [0.3977529836363989], "n_samples_seen_": 55}}, {"input_preprocessor": {"num": {"mean_": [1.6307636010361506, 75.75535714285715, 832.615480014256], "var_": [0.13706793334467066, 768.2367570153064, 385465.5730380972], "scale_": [0.3702268673998021, 27.717084208395846, 620.8587383923151], "n_samples_seen_": 56}}, "output_preprocessor": {"mean_": [0.44402859465169187], "var_": [0.16073403177350493], "scale_": [0.40091648977499655], "n_samples_seen_": 56}}]}, "extras": [{}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}]}}