Merge branch 'master' of https://github.com/ddobric/neocortexapi

ddobric · Jan 16, 2021 · 4ab44bb · 4ab44bb
2 parents c6f3830 + 8f05c71
commit 4ab44bb
Show file tree

Hide file tree

Showing 35 changed files with 1,116 additions and 368 deletions.
diff --git a/NeoCortexApi/AkkaSb.Net/AkkaSb.Net.csproj b/NeoCortexApi/AkkaSb.Net/AkkaSb.Net.csproj
@@ -9,6 +9,13 @@
 
   <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|AnyCPU'">
     <GenerateSerializationAssemblies>Auto</GenerateSerializationAssemblies>
+    <DebugType>full</DebugType>
+    <DebugSymbols>true</DebugSymbols>
+  </PropertyGroup>
+
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|AnyCPU'">
+    <DebugType>full</DebugType>
+    <DebugSymbols>true</DebugSymbols>
   </PropertyGroup>
 
   <ItemGroup>

diff --git a/NeoCortexApi/DistributedComputeLib/DistributedComputeLib.csproj b/NeoCortexApi/DistributedComputeLib/DistributedComputeLib.csproj
@@ -6,6 +6,7 @@
   </PropertyGroup>
 
   <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|AnyCPU'">
+    <DocumentationFile></DocumentationFile>
   </PropertyGroup>
 
   <ItemGroup>

diff --git a/NeoCortexApi/Documentation/how-does-htmclassifier-work.md b/NeoCortexApi/Documentation/how-does-htmclassifier-work.md
@@ -0,0 +1,93 @@
+## HtmClassifier
+The HtmClassifier is a helper module that is used to help the process of the next element in the process of learning sequences.
+The classifier provides two methods:
+
+Learn(string key, int[] sdr)
+
+String Predict(int[] predictiveCells)
+
+The method learn receives the key string, that represents the sequence and memorizes the SDR for the given key.
+Assume, we learn following sequence: 
+~~~
+1-2-3-4-5-3-5
+~~~
+
+In every cycle, the experiment creates the key that represents the sequence in that cycle. For example, the key might look like:
+
+Cycle 1: '1-2-3-4-5-3-5' , 
+Cycle 2: '2-3-4-5-3-5-1', 
+Cycle 3: '3-4-5-3-5-1-2', 
+etc..
+
+During the learning process, the input in every cycle is SDR of cells produced by Temporal Memory algorithm. Because the same SP output (column SDR) for some element (i.e.: ‘3’) will be represented in TM by a different set of cells inside of the same column set. SP generates always (if stable) the same set of active columns for the same element. However, TM does not generate the same set of active cells for the same element. The TM is trying to build the context of the element.
+That means ‘3’ followed by ‘2’ produces a different set of active cells than ‘3’ followed by ‘5’. This is why the classifier gets the key in the form shown above. However, developers are free to build a key some other way.
+
+The following shows the trace output of the learning process.
+
+~~~
+Col  SDR: 3, 451, 515, 532, 534, 972, 976, 979, 981, 984, 986, 997, 1005, 1013, 1014, 1015, 1019, 1020, 1021, 1022, 
+Cell SDR: *94, 11287, 12895, 13312, 13370, 24302, 24402, 24479, 24542, 24609, 24666, 24925, 25132, 25342, 25354, 25375, 25477, 25513, 25526, 25560,*
+Missmatch! Actual value: 14-5-7-6-9-3-4-3-4-3-4-0-1-0-2-3-4-5-6-5-4-3-7-1-9-12-11-12-13-14-11-12 - Predicted value: 
+Item length: 16	 Items: 34
+Predictive cells: 16 	 530, 855, 1213, 1228, 1339, 1988, 2318, 2925, 13843, 14641, 14961, 15043, 15322, 24538, 24932, 25268, 
+<indx:0	inp/len: -1.0-0-1-0-2-3-4-5-6-5-4-3-7-1-9-12-11-12-13-14-11-12-14-5-7-6-9-3-4-3-4/20 = similarity 0	 2475, 3145, 3164, 3351, 3674, 3887, 3935, 4329, 14118, 15078, 15514, 15746, 16174, 16329, 16426, 16704, 16837, 17311, 17368, 17429, 
+<indx:1	inp/len: -1.0-0-1-0-2-3-4-5-6-5-4-3-7-1-9-12-11-12-13-14-11-12-14-5-7-6-9-3-4-3-4-3/20 = similarity 0	 2378, 2490, 2584, 3079, 3106, 3137, 3158, 3230, 3363, 3838, 13649, 13960, 14122, 14249, 14288, 14992, 15098, 15587, 16344, 16543, 
+<indx:2	inp/len: 0-1-0-2-3-4-5-6-5-4-3-7-1-9-12-11-12-13-14-11-12-14-5-7-6-9-3-4-3-4-3-4/40 = similarity 0	 2475, 2495, 3145, 3148, 3152, 3164, 3351, 3372, 3663, 3674, 3887, 3894, 3935, 3937, 4328, 4329, 14101, 14118, 15078, 15086, 15514, 15520, 15742, 15746, 16156, 16174, 16329, 16336, 16426, 16449, 16704, 16708, 16837, 16838, 17300, 17311, 17366, 17368, 17429, 17445, 
+<indx:3	inp/len: 1-0-2-3-4-5-6-5-4-3-7-1-9-12-11-12-13-14-11-12-14-5-7-6-9-3-4-3-4-3-4-0/20 = similarity 0	 10403, 10499, 10671, 10920, 11028, 11481, 12200, 12475, 12544, 12762, 13030, 13121, 13273, 13463, 22984, 23277, 23979, 24388, 24733, 25288, 
+<indx:4	inp/len: 0-2-3-4-5-6-5-4-3-7-1-9-12-11-12-13-14-11-12-14-5-7-6-9-3-4-3-4-3-4-0-1/20 = similarity 0	 445, 10487, 11026, 11478, 11562, 11627, 12009, 12064, 12214, 12500, 12528, 13032, 13110, 13264, 13464, 13617, 13898, 13940, 14319, 25576, 
+<indx:5	inp/len: 2-3-4-5-6-5-4-3-7-1-9-12-11-12-13-14-11-12-14-5-7-6-9-3-4-3-4-3-4-0-1-0/20 = similarity 0	 10401, 10492, 10661, 10924, 11047, 11478, 12220, 12484, 12542, 12752, 13027, 13107, 13260, 13453, 22987, 23281, 23983, 24380, 24734, 25279, 
+<indx:6	inp/len: 3-4-5-6-5-4-3-7-1-9-12-11-12-13-14-11-12-14-5-7-6-9-3-4-3-4-3-4-0-1-0-2/20 = similarity 0	 445, 2102, 2683, 11478, 11562, 12812, 13617, 13629, 13807, 13898, 14243, 14280, 14512, 14586, 14944, 15131, 15168, 15203, 15282, 15578, 
+<indx:7	inp/len: 4-5-6-5-4-3-7-1-9-12-11-12-13-14-11-12-14-5-7-6-9-3-4-3-4-3-4-0-1-0-2-3/20 = similarity 0	 2384, 2479, 2586, 3081, 3118, 3139, 3173, 3231, 3371, 3836, 13641, 13952, 14114, 14233, 14289, 14999, 15093, 15576, 16341, 16538, 
+<indx:8	inp/len: 5-6-5-4-3-7-1-9-12-11-12-13-14-11-12-14-5-7-6-9-3-4-3-4-3-4-0-1-0-2-3-4/20 = similarity 0	 2475, 3145, 3164, 3351, 3674, 3887, 3935, 4329, 14118, 15078, 15514, 15746, 16174, 16329, 16426, 16704, 16837, 17311, 17368, 17429, 
+<indx:9	inp/len: 6-5-4-3-7-1-9-12-11-12-13-14-11-12-14-5-7-6-9-3-4-3-4-3-4-0-1-0-2-3-4-5/20 = similarity 0	 5010, 5214, 16405, 16590, 16781, 16832, 16933, 17202, 17440, 17466, 17556, 17700, 17790, 17815, 17973, 18045, 18249, 18352, 18396, 18486, 
+<indx:10	inp/len: 5-4-3-7-1-9-12-11-12-13-14-11-12-14-5-7-6-9-3-4-3-4-3-4-0-1-0-2-3-4-5-6/20 = similarity 0	 17201, 17449, 17721, 17817, 18243, 18296, 18354, 18423, 18828, 18948, 19043, 19271, 19324, 19388, 19529, 19568, 19598, 19679, 19794, 19854, 
+<indx:11	inp/len: 4-3-7-1-9-12-11-12-13-14-11-12-14-5-7-6-9-3-4-3-4-3-4-0-1-0-2-3-4-5-6-5/20 = similarity 0	 5015, 5224, 16415, 16587, 16777, 16849, 16946, 17218, 17438, 17474, 17555, 17720, 17791, 17801, 17968, 18028, 18226, 18374, 18389, 18480, 
+<indx:12	inp/len: 3-7-1-9-12-11-12-13-14-11-12-14-5-7-6-9-3-4-3-4-3-4-0-1-0-2-3-4-5-6-5-4/20 = similarity 0	 2478, 3127, 3162, 3370, 3652, 3890, 3932, 4337, 14119, 15089, 15517, 15741, 16169, 16342, 16443, 16705, 16835, 17315, 17352, 17435, 
+<indx:13	inp/len: 7-1-9-12-11-12-13-14-11-12-14-5-7-6-9-3-4-3-4-3-4-0-1-0-2-3-4-5-6-5-4-3/40 = similarity 0	 2378, 2386, 2482, 2490, 2583, 2584, 3079, 3093, 3106, 3113, 3137, 3140, 3158, 3166, 3230, 3237, 3363, 3365, 3837, 3838, 13628, 13649, 13957, 13960, 14115, 14122, 14227, 14249, 14288, 14298, 14992, 14998, 15082, 15098, 15587, 15596, 16325, 16344, 16534, 16543, 
+<indx:14	inp/len: 1-9-12-11-12-13-14-11-12-14-5-7-6-9-3-4-3-4-3-4-0-1-0-2-3-4-5-6-5-4-3-7/20 = similarity 0	 6936, 7061, 18236, 18367, 18542, 18587, 18803, 18949, 18950, 19026, 19310, 19537, 19551, 19692, 19782, 20058, 20179, 20332, 20452, 20853, 
+<indx:15	inp/len: 9-12-11-12-13-14-11-12-14-5-7-6-9-3-4-3-4-3-4-0-1-0-2-3-4-5-6-5-4-3-7-1/20 = similarity 0	 435, 10480, 11027, 11498, 11568, 11643, 12021, 12074, 12223, 12512, 12532, 13044, 13120, 13274, 13455, 13613, 13892, 13949, 14303, 25577, 
+<indx:16	inp/len: 12-11-12-13-14-11-12-14-5-7-6-9-3-4-3-4-3-4-0-1-0-2-3-4-5-6-5-4-3-7-1-9/20 = similarity 0	 9485, 9518, 9616, 9648, 20404, 20803, 20944, 21044, 21122, 21469, 21531, 21737, 21910, 22232, 22387, 22445, 22540, 22603, 22642, 22896, 
+<indx:17	inp/len: 11-12-13-14-11-12-14-5-7-6-9-3-4-3-4-3-4-0-1-0-2-3-4-5-6-5-4-3-7-1-9-12/20 = similarity 0	 92, 11281, 12884, 13310, 13361, 24303, 24400, 24489, 24541, 24612, 24654, 24947, 25139, 25349, 25364, 25387, 25476, 25522, 25535, 25574, 
+>indx:18	inp/len: 12-13-14-11-12-14-5-7-6-9-3-4-3-4-3-4-0-1-0-2-3-4-5-6-5-4-3-7-1-9-12-11/20 = similarity 2	 11532, 11734, 11998, 21032, 22607, 22790, 22888, 23081, 23120, 23361, 23554, 23939, 24011, 24258, 24418, 24486, 24538, 24620, 24673, 24932, 
+<indx:19	inp/len: 13-14-11-12-14-5-7-6-9-3-4-3-4-3-4-0-1-0-2-3-4-5-6-5-4-3-7-1-9-12-11-12/20 = similarity 0	 91, 11296, 12875, 13311, 13354, 24322, 24405, 24480, 24544, 24606, 24656, 24935, 25148, 25343, 25352, 25391, 25496, 25521, 25541, 25572, 
+<indx:20	inp/len: 14-11-12-14-5-7-6-9-3-4-3-4-3-4-0-1-0-2-3-4-5-6-5-4-3-7-1-9-12-11-12-13/20 = similarity 0	 890, 1063, 1153, 12869, 12893, 13142, 13364, 13748, 13767, 13911, 24159, 24539, 24948, 25125, 25335, 25365, 25479, 25505, 25527, 25571, 
+*** >indx:21	inp/len: 11-12-14-5-7-6-9-3-4-3-4-3-4-0-1-0-2-3-4-5-6-5-4-3-7-1-9-12-11-12-13-14/20 = similarity 14	 530, 855, 1063, 1153, 1213, 1228, 1339, 1988, 2318, 2925, 13843, 14049, 14188, 14641, 14961, 15043, 15322, 25268, 25365, 25571, 
+<indx:22	inp/len: 12-14-5-7-6-9-3-4-3-4-3-4-0-1-0-2-3-4-5-6-5-4-3-7-1-9-12-11-12-13-14-11/20 = similarity 0	 11535, 11748, 11987, 21046, 22622, 22784, 22882, 23088, 23109, 23355, 23565, 23930, 24007, 24262, 24409, 24490, 24526, 24619, 24664, 24936, 
+<indx:23	inp/len: 14-5-7-6-9-3-4-3-4-3-4-0-1-0-2-3-4-5-6-5-4-3-7-1-9-12-11-12-13-14-11-12/20 = similarity 0	 *94, 11287, 12895, 13312, 13370, 24302, 24402, 24479, 24542, 24609, 24666, 24925, 25132, 25342, 25354, 25375, 25477, 25513, 25526, 25560,* 
+*** <indx:24	inp/len:>5-7-6-9-3-4-3-4-3-4-0-1-0-2-3-4-5-6-5-4-3-7-1-9-12-11-12-13-14-11-12-14/20 = similarity 14	 530, 855, 1063, 1153, 1213, 1228, 1339, 1988, 2318, 2925, 13843, 14037, 14191, 14641, 14961, 15043, 15322, 25268, 25365, 25571, 
+<indx:25	inp/len: 7-6-9-3-4-3-4-3-4-0-1-0-2-3-4-5-6-5-4-3-7-1-9-12-11-12-13-14-11-12-14-5/20 = similarity 0	 5017, 5209, 16420, 16596, 16789, 16843, 16943, 17217, 17439, 17467, 17574, 17724, 17798, 17803, 17958, 18026, 18232, 18357, 18387, 18490, 
+<indx:26	inp/len: 6-9-3-4-3-4-3-4-0-1-0-2-3-4-5-6-5-4-3-7-1-9-12-11-12-13-14-11-12-14-5-7/20 = similarity 0	 6933, 7062, 18237, 18364, 18533, 18589, 18824, 18932, 18965, 19048, 19317, 19540, 19570, 19699, 19780, 20055, 20178, 20325, 20455, 20867, 
+<indx:27	inp/len: 9-3-4-3-4-3-4-0-1-0-2-3-4-5-6-5-4-3-7-1-9-12-11-12-13-14-11-12-14-5-7-6/20 = similarity 0	 17218, 17438, 17720, 17801, 18226, 18291, 18374, 18408, 18834, 18925, 19049, 19266, 19323, 19378, 19536, 19557, 19588, 19696, 19781, 19863, 
+<indx:28	inp/len: 3-4-3-4-3-4-0-1-0-2-3-4-5-6-5-4-3-7-1-9-12-11-12-13-14-11-12-14-5-7-6-9/20 = similarity 0	 9497, 9506, 9620, 9633, 20423, 20806, 20949, 21045, 21104, 21450, 21526, 21734, 21909, 22246, 22389, 22442, 22526, 22613, 22636, 22876, 
+<indx:29	inp/len: 4-3-4-3-4-0-1-0-2-3-4-5-6-5-4-3-7-1-9-12-11-12-13-14-11-12-14-5-7-6-9-3/20 = similarity 0	 2388, 2483, 2585, 3087, 3122, 3141, 3169, 3243, 3352, 3828, 13648, 13973, 14123, 14231, 14281, 14987, 15090, 15589, 16335, 16530, 
+<indx:30	inp/len: 3-4-3-4-0-1-0-2-3-4-5-6-5-4-3-7-1-9-12-11-12-13-14-11-12-14-5-7-6-9-3-4/20 = similarity 0	 2489, 3131, 3170, 3358, 3669, 3896, 3927, 4332, 14104, 15075, 15501, 15745, 16167, 16327, 16446, 16715, 16826, 17307, 17350, 17446, 
+<indx:31	inp/len: 4-3-4-0-1-0-2-3-4-5-6-5-4-3-7-1-9-12-11-12-13-14-11-12-14-5-7-6-9-3-4-3/40 = similarity 0	 2378, 2386, 2482, 2490, 2583, 2584, 3079, 3093, 3106, 3113, 3137, 3140, 3158, 3166, 3230, 3237, 3363, 3365, 3837, 3838, 13628, 13649, 13957, 13960, 14115, 14122, 14227, 14249, 14288, 14298, 14992, 14998, 15082, 15098, 15587, 15596, 16325, 16344, 16534, 16543, 
+<indx:32	inp/len: 3-4-0-1-0-2-3-4-5-6-5-4-3-7-1-9-12-11-12-13-14-11-12-14-5-7-6-9-3-4-3-4/20 = similarity 0	 2495, 3148, 3152, 3372, 3663, 3894, 3937, 4328, 14101, 15086, 15520, 15742, 16156, 16336, 16449, 16708, 16838, 17300, 17366, 17445, 
+<indx:33	inp/len: 4-0-1-0-2-3-4-5-6-5-4-3-7-1-9-12-11-12-13-14-11-12-14-5-7-6-9-3-4-3-4-3/20 = similarity 0	 2376, 2496, 2595, 3084, 3103, 3146, 3161, 3242, 3374, 3849, 13630, 13972, 14112, 14247, 14290, 14997, 15081, 15593, 16340, 16531, 
+Current Input: 12 	| Predicted Input: *11-12-14-5-7-6-9-3-4-3-4-3-4-0-1-0-2-3-4-5-6-5-4-3-7-1-9-12-11-12-13-14*
+
+-------------- 14 --------------- NEXT CYCLE
+PPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPP
+Active segments: 40, Matching segments: 40
+Col  SDR: 21, 34, 42, 46, 48, 49, 53, 79, 92, 117, 553, 561, 567, 585, 598, 601, 612, 1010, 1014, 1022, 
+Cell SDR: 530, 855, 1063, 1153, 1213, 1228, 1339, 1988, 2318, 2925, 13843, 14037, 14191, 14641, 14961, 15043, 15322, 25268, 25365, 25571, 
+Missmatch! Actual value: 5-7-6-9-3-4-3-4-3-4-0-1-0-2-3-4-5-6-5-4-3-7-1-9-12-11-12-13-14-11-12-14 - Predicted value: 11-12-14-5-7-6-9-3-4-3-4-3-4-0-1-0-2-3-4-5-6-5-4-3-7-1-9-12-11-12-13-14
+~~~
+
+How to read the trace? It is very simple. The trace shows the memorized SDRs with the matching score 'similarity'.
+In the current cycle following SDR was observed:
+Cell SDR: 94, 11287, 12895, 13312, 13370, 24302, 24402, 24479, 24542, 24609, 24666, 24925, 25132, 25342, 25354, 25375, 25477, 25513, 25526, 25560,
+
+The classifier is traversing through all memorized SDRs and tries to match the best one. In this case, there are two SDRs with the 14 matching cells at index 21 and 24.
+The first one represents the sequence 
+~~~
+11-12-14-5-7-6-9-3-4-3-4-3-4-0-1-0-2-3-4-5-6-5-4-3-7-1-9-12-11-12-13-14
+~~~ 
+and the second one 
+~~~
+5-7-6-9-3-4-3-4-3-4-0-1-0-2-3-4-5-6-5-4-3-7-1-9-12-11-12-13-14-11-12-14
+~~~
+
+The current implementation of the classifier peeks the first best matching one, which is in more complex sequences not a sufficient solution. As you see, the '14' was correcttly predicted. But '14' after '13' is a different context than '14' after '12'.
+
+We are considering to improve the classifier to be able to detect more complex sequence.
diff --git a/NeoCortexApi/HtmAkkaHost/HtmAkkaHost.csproj b/NeoCortexApi/HtmAkkaHost/HtmAkkaHost.csproj
@@ -6,6 +6,16 @@
     <DockerDefaultTargetOS>Linux</DockerDefaultTargetOS>
   </PropertyGroup>
 
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|AnyCPU'">
+    <DebugType>full</DebugType>
+    <DebugSymbols>true</DebugSymbols>
+  </PropertyGroup>
+
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|AnyCPU'">
+    <DebugType>full</DebugType>
+    <DebugSymbols>true</DebugSymbols>
+  </PropertyGroup>
+
   <ItemGroup>
     <PackageReference Include="Akka.Remote" Version="1.3.13" />
     <PackageReference Include="Microsoft.Extensions.Configuration" Version="2.2.0" />

diff --git a/NeoCortexApi/NeoCortexApi.xml b/NeoCortexApi/NeoCortexApi.xml
diff --git a/NeoCortexApi/NeoCortexApi/HomeostaticPlasticityController.cs b/NeoCortexApi/NeoCortexApi/HomeostaticPlasticityController.cs
@@ -79,6 +79,14 @@ protected HomeostaticPlasticityController()
 
         }
 
+        /// <summary>
+        /// Creates the instance of HomeostaticPlasticityController.
+        /// </summary>
+        /// <param name="htmMemory">The HTM memory.</param>
+        /// <param name="minCycles">The minimum calls to the Learn method until HPC algorithm is activated. When this number is reached the HPC will disable boosting in SP. </param>
+        /// <param name="onStabilityStatusChanged">Action invoked when the SP status is changed from stable t unstable and vise versa.</param>
+        /// <param name="numOfCyclesToWaitOnChange">How many cycles all seen patterns must not change to declare SP as stable. Using smaller numbers might cause frequent status change.
+        /// Higher numbers ensure more stable SP, but it takes longer time to enter the stable stabe.</param>
         public HomeostaticPlasticityController(Connections htmMemory, int minCycles, Action<bool, int, double, int> onStabilityStatusChanged, int numOfCyclesToWaitOnChange = 50)
         {
             this.m_OnStabilityStatusChanged = onStabilityStatusChanged;
@@ -407,7 +415,7 @@ public void Serialize(StreamWriter writer)
             ser.SerializeBegin(nameof(HomeostaticPlasticityController), writer);
 
             ser.SerializeValue(this.m_MaxPreviousElements, writer);
-            // HtmMemory is not serialized here. It is assumed to be serialized in the SP;
+            // m_HtmMemory is not serialized here. It is assumed to be serialized in the SP;
             ser.SerializeValue(this.m_Cycle, writer);
             ser.SerializeValue(this.m_MinCycles, writer);
             ser.SerializeValue(this.m_RequiredNumOfStableCycles, writer);
@@ -425,10 +433,10 @@ public static HomeostaticPlasticityController Deserialize(StreamReader reader)
             HomeostaticPlasticityController ctrl = new HomeostaticPlasticityController();
 
             HtmSerializer2 ser = new HtmSerializer2();
-            ctrl.m_MaxPreviousElements = ser.ReadIntValue(reader);
-            ctrl.m_MinCycles = ser.ReadIntValue(reader);
-            ctrl.m_RequiredNumOfStableCycles = ser.ReadIntValue(reader);
-            //...
+            //ctrl.m_MaxPreviousElements = ser.ReadIntValue(reader);
+            //ctrl.m_MinCycles = ser.ReadIntValue(reader);
+            //ctrl.m_RequiredNumOfStableCycles = ser.ReadIntValue(reader);
+            ////...
 
             return ctrl;
 

diff --git a/NeoCortexApi/NeoCortexApi/NeoCortexApi.csproj b/NeoCortexApi/NeoCortexApi/NeoCortexApi.csproj
@@ -25,6 +25,13 @@
   <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|AnyCPU'">
     <LangVersion>7.3</LangVersion>
     <DefineConstants>TRACE;USE_AKKA</DefineConstants>
+    <DebugType>full</DebugType>
+    <DebugSymbols>true</DebugSymbols>
+  </PropertyGroup>
+
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|AnyCPU'">
+    <DebugType>full</DebugType>
+    <DebugSymbols>true</DebugSymbols>
   </PropertyGroup>
 
   <PropertyGroup Condition=" '$(Configuration)|$(Platform)' == 'Release|AnyCPU' ">
@@ -61,6 +68,9 @@
   </ItemGroup>
 
   <ItemGroup>
+	  <ProjectReference Include="..\AkkaSb.Net\AkkaSb.Net.csproj" >
+      <PrivateAssets>all</PrivateAssets>
+    </ProjectReference>
 	  <ProjectReference Include="..\DistributedComputeLib\DistributedComputeLib.csproj">
 		  <PrivateAssets>all</PrivateAssets>
 	  </ProjectReference>