-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathsources.txt
112 lines (85 loc) · 3.15 KB
/
sources.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
66 ./src/main/scala/qd/Main.scala
34 ./src/main/scala/qd/Lineage.scala
142 ./src/main/scala/qd/value.scala
66 ./src/main/scala/qd/TokenVec.scala
114 ./src/main/scala/qd/base.scala
109 ./src/main/scala/qd/instance/Instance.scala
39 ./src/main/scala/qd/instance/Assignment.scala
179 ./src/main/scala/qd/instance/AssignmentTrie.scala
33 ./src/main/scala/qd/instance/Config.scala
15 ./src/main/scala/qd/evaluator/Evaluator.scala
85 ./src/main/scala/qd/evaluator/NaiveEvaluator.scala
111 ./src/main/scala/qd/evaluator/SeminaiveEvaluator.scala
136 ./src/main/scala/qd/evaluator/TrieEvaluator.scala
123 ./src/main/scala/qd/evaluator/TrieSemiEvaluator.scala
100 ./src/main/scala/qd/Enumerator.scala
108 ./src/main/scala/qd/problem/Problem.scala
248 ./src/main/scala/qd/problem/QDParser.scala
105 ./src/main/scala/qd/learner/Scorer.scala
127 ./src/main/scala/qd/learner/Learner.scala
-----
60 ./src/test/scala/qd/BaseSpec.scala
104 ./src/test/scala/qd/instance/InstanceSpec.scala
84 ./src/test/scala/qd/instance/AssignmentTrieSpec.scala
92 ./src/test/scala/qd/data/graphs/Graphs.scala
29 ./src/test/scala/qd/data/graphs/Reachability.scala
21 ./src/test/scala/qd/evaluator/EvaluatorReachabilitySpec.scala
70 ./src/test/scala/qd/EnumeratorSpec.scala
172 ./src/test/scala/qd/ParserSpec.scala
43 ./src/test/scala/qd/learner/LearnerSpec.scala
-----
108 ./src/main/scala/qd/util/bdd/BDD.scala
34 ./src/main/scala/qd/util/Multiset.scala
61 ./src/test/scala/qd/util/MultisetSpec.scala
-----
2818 total
--------------------------------------------------------------------------------
0. Experiment Scripts
1. Cleanup
a. Choose appropriate collections: Vector, List, Seq
b. Map.mapValues
c. lazy val
d. hashCode
2. Optimization
a. Cross entropy
b. Gradient descent
c. Parallel runs
3. Training Details
a. Statistical significance (t-test)
4. New Benchmarks
a. Kinship, Nations, UMLS
c. FB15K, WN18
5. Better join algorithms
a. Assignments as tries
--------------------------------------------------------------------------------
0. Title + Abstract
1. Introduction
2. Related Work
3. Framework
a. Introduction to Datalog
b. Program Learning as Rule Selection
- NP-completeness of rule selection
c. Difflog
- Choice of semiring
- Datalog forms a Galois connection with Difflog
- PTIME-completeness of Difflog
- Learning by Continuous Optimization + Details of Loss Function
d. Optimizations for Inference
- Trie-based evaluator
- Join algorithms
- Valuation collapse
4. Experiments
- Protocol, optimization parameters
- Statistical significance of results
- Learning Combinatorial Programs
Vs. ALPS, Metagol, Zaatar. Synthesis time, synthesis accuracy
Predicate invention
- Knowledge-Base Completion
Formalizing the KBC problem
Highlight possibility of relationship repurposing
Vs. Rocktaschel, Cohen, 2014
Countries S1, S2, S3, Kinship, Nations, UMLS
FB15 WN18
5. Future Directions
- Better join algorithms / TensorFlow
- Combining CDCL and continuous optimization