Open Hub
Open Hub

 Settings |  Report Duplicate

0
I Use This!
Very Low Activity

Commits : Listings

Analyzed about 4 hours ago. based on code collected 1 day ago.
Nov 05, 2025 — Nov 05, 2026
Commit Message Contributor Files Modified Lines Added Lines Removed Code Location Date
------------------------------------------------------------------------------------------------------ Support Vector Machines Update and Discrete Hyperbolic Factorization Spark Benchmarks ------------------------------------------------------------------------------------------------------ (*) Support Vector Machines implementation has been updated to persist the learnt support vectors to a disk file SupportVectorMachines.txt (*) This text file is read in classify() (*) Support Vectors Dictionary is JSON dumped and eval()-ed and not JSON loaded because of serialization glitch in defaultdict(list) ------------------------------------------ (*) Computational Geometric Hyperbolic Factorization Spark implementation has been benchmarked for 2 more integers of 9 and 10 bits. Some 8 bit numbers were benchmarked earlier. (*) Spark logs for these benchmarks have been committed and time duration is calculated as 3 way split of real/user/system by time shell utility (*) These benchmark numbers are on dual core single node Spark cluster (*) C++/python files for tilings have been updated and pixelated tiles storage text files have been rewritten
Srinivasan Kannan
as K.Srinivasan
More... almost 9 years ago
Updated AsFer Design Document
Srinivasan Kannan
as K.Srinivasan
More... almost 9 years ago
Updated AsFer Design Document
Srinivasan Kannan
as K.Srinivasan
More... almost 9 years ago
------------------------------------------------------------------------------------------------------------------------------------ 469. (FEATURE-DONE) Computational Geometric Hyperbolic Factorization - Pixelated Segments Spark Bitonic Sort - 13 November 2017 ------------------------------------------------------------------------------------------------------------------------------------ (*) Number to Factorize has been increased to 875 from 147 (*) C++ tiling pre-processing routines have been changed for this and Pixelated Tiles storage text files for bitonic sort have been updated (*) Factorization of 16321 is benchmarked on single node cluster on dual core (which is parallel RAM). (*) This is just a representative number on single dual-core CPU and not a cloud parallelism benchmark. (*) Each DAGScheduler Spark work item is independent code executable in parallel and benchmark has to be this parallel time per work unit which is captured by per task duration in Spark Driver logs below: 17/11/13 21:18:22 INFO TaskSetManager: Finished task 1.0 in stage 1791.0 (TID 3583) in 46 ms on localhost (executor driver) (1/2) 17/11/13 21:18:22 INFO TaskSetManager: Finished task 0.0 in stage 1791.0 (TID 3582) in 47 ms on localhost (executor driver) (2/2) 17/11/13 21:18:22 INFO TaskSetManager: Starting task 0.0 in stage 1792.0 (TID 3584, localhost, executor driver, partition 0, PROCESS_LOCAL, 6281 bytes) 17/11/13 21:18:22 INFO TaskSetManager: Starting task 1.0 in stage 1792.0 (TID 3585, localhost, executor driver, partition 1, PROCESS_LOCAL, 6309 bytes) 17/11/13 21:18:22 INFO TaskSetManager: Finished task 1.0 in stage 1792.0 (TID 3585) in 46 ms on localhost (executor driver) (1/2) 17/11/13 21:18:22 INFO TaskSetManager: Finished task 0.0 in stage 1792.0 (TID 3584) in 50 ms on localhost (executor driver) (2/2) 17/11/13 21:26:39 INFO TaskSetManager: Finished task 0.0 in stage 1791.0 (TID 3582) in 50 ms on localhost (executor driver) (1/2) 17/11/13 21:26:39 INFO TaskSetManager: Finished task 1.0 in stage 1791.0 (TID 3583) in 49 ms on localhost (executor driver) (2/2) 17/11/13 21:26:39 INFO TaskSetManager: Starting task 0.0 in stage 1792.0 (TID 3584, localhost, executor driver, partition 0, PROCESS_LOCAL, 6281 bytes) 17/11/13 21:26:39 INFO TaskSetManager: Starting task 1.0 in stage 1792.0 (TID 3585, localhost, executor driver, partition 1, PROCESS_LOCAL, 6309 bytes) 17/11/13 21:26:39 INFO TaskSetManager: Finished task 0.0 in stage 1792.0 (TID 3584) in 48 ms on localhost (executor driver) (1/2) 17/11/13 21:26:39 INFO TaskSetManager: Finished task 1.0 in stage 1792.0 (TID 3585) in 49 ms on localhost (executor driver) (2/2) 17/11/13 21:33:59 INFO TaskSetManager: Finished task 0.0 in stage 1791.0 (TID 3582) in 48 ms on localhost (executor driver) (1/2) 17/11/13 21:33:59 INFO TaskSetManager: Finished task 1.0 in stage 1791.0 (TID 3583) in 50 ms on localhost (executor driver) (2/2) 17/11/13 21:33:59 INFO TaskSetManager: Starting task 0.0 in stage 1792.0 (TID 3584, localhost, executor driver, partition 0, PROCESS_LOCAL, 6281 bytes) 17/11/13 21:33:59 INFO TaskSetManager: Starting task 1.0 in stage 1792.0 (TID 3585, localhost, executor driver, partition 1, PROCESS_LOCAL, 6309 bytes) 17/11/13 21:33:59 INFO TaskSetManager: Finished task 0.0 in stage 1792.0 (TID 3584) in 55 ms on localhost (executor driver) (1/2) 17/11/13 21:33:59 INFO TaskSetManager: Finished task 1.0 in stage 1792.0 (TID 3585) in 57 ms on localhost (executor driver) (2/2) 17/11/13 21:40:20 INFO TaskSetManager: Finished task 1.0 in stage 1791.0 (TID 3583) in 49 ms on localhost (executor driver) (1/2) 17/11/13 21:40:20 INFO TaskSetManager: Finished task 0.0 in stage 1791.0 (TID 3582) in 52 ms on localhost (executor driver) (2/2) 17/11/13 21:40:20 INFO TaskSetManager: Starting task 0.0 in stage 1792.0 (TID 3584, localhost, executor driver, partition 0, PROCESS_LOCAL, 6281 bytes) 17/11/13 21:40:20 INFO TaskSetManager: Starting task 1.0 in stage 1792.0 (TID 3585, localhost, executor driver, partition 1, PROCESS_LOCAL, 6309 bytes) 17/11/13 21:40:20 INFO TaskSetManager: Finished task 0.0 in stage 1792.0 (TID 3584) in 50 ms on localhost (executor driver) (1/2) 17/11/13 21:40:20 INFO TaskSetManager: Finished task 1.0 in stage 1792.0 (TID 3585) in 50 ms on localhost (executor driver) (2/2) (*) Bitonic Sort is O(logN*logN) and number of processors required is approximately O(N^2.5) but abides by NC definition. Cole Pipelined Merge Sort which is O(logN) is both in NC and Work-Time Optimal.
Srinivasan Kannan
as K.Srinivasan
More... almost 9 years ago
Updated AsFer Design Document
Srinivasan Kannan
as K.Srinivasan
More... almost 9 years ago
Updated AsFer Design Document
Srinivasan Kannan
as K.Srinivasan
More... almost 9 years ago
Updated AsFer Design Document
Srinivasan Kannan
as K.Srinivasan
More... almost 9 years ago
Updated AsFer Design Document
Srinivasan Kannan
as K.Srinivasan
More... almost 9 years ago
More Logs for SupportVectorMachines.py
Srinivasan Kannan
More... almost 9 years ago
Updated AsFer Design Document
Srinivasan Kannan
as K.Srinivasan
More... almost 9 years ago
--------------------------------------------------------------------------------------------------- Support Vector Machines - update --------------------------------------------------------------------------------------------------- (*) Numpy indexing has been changed (*) Both random point and parametric point distances have been tested (*) DCCP log and Support Vector logs for 2 points and a random point have been committed to testlogs/ (*) Distances are printed in the matrix result - two diametrically opposite points have equal distances (*) Distance matrix is returned from distance_from_separating_hyperplane() function (*) The distance minimized is the L1 norm (sum of tuple elements) and not L2 norm (sum of squares of tuple elements)
Srinivasan Kannan
as K.Srinivasan
More... almost 9 years ago
Updated AsFer Design Document
Srinivasan Kannan
as K.Srinivasan
More... almost 9 years ago
Updated AsFer Design Document
Srinivasan Kannan
as K.Srinivasan
More... almost 9 years ago
------------------------------------------------------------------------------------------------------ Support Vector Machines implementation - based on CVXPY ------------------------------------------------------------------------------------------------------ (*) New Support Vector Machines python implementation is committed to NeuronRain AsFer repository (*) This minimizes and objective function 1/2*||w|| subject to constraint ||WX+b|| >= 1 which labels a point +1 or -1 on either side of a decision separating hyperplane WX+b for weight vector W and bias b (Reference: Machine Learning - Ethem Alpaydin - Chapter 13 - Kernel Machines) (*) Optimization is solved by CVXPY DCCP Convex-Concave program (*) logs for this have been added to testlogs/ (*) Present NeuronRain AsFer SVMRetriever.cpp depends on third-party SVMlight opensource software. With this new implementation, references to SVMlight are to be phased out.
Srinivasan Kannan
as K.Srinivasan
More... almost 9 years ago
Updated AsFer Design Document
Srinivasan Kannan
as K.Srinivasan
More... almost 9 years ago
Updated AsFer Design Document
Srinivasan Kannan
as K.Srinivasan
More... almost 9 years ago
Updated AsFer Design Document
Srinivasan Kannan
as K.Srinivasan
More... almost 9 years ago
Updated AsFer Design Document
Srinivasan Kannan
as K.Srinivasan
More... about 9 years ago
Updated AsFer Design Document
Srinivasan Kannan
as K.Srinivasan
More... about 9 years ago
Updated AsFer Design Document
Srinivasan Kannan
as K.Srinivasan
More... about 9 years ago
Updated AsFer Design Document
Srinivasan Kannan
as K.Srinivasan
More... about 9 years ago
Updated AsFer Design Document
Srinivasan Kannan
as K.Srinivasan
More... about 9 years ago
Updated AsFer Design Document
Srinivasan Kannan
as K.Srinivasan
More... about 9 years ago
Updated AsFer Design Document
Srinivasan Kannan
as K.Srinivasan
More... about 9 years ago
----------------------------------------------------------------------------------------------- Ephemeris Search - Sequence Mining of Tropical Monsoon for mid-November 2017 - 1 November 2017 ----------------------------------------------------------------------------------------------- (#) Apriori GSP SequenceMining on autogen_classifier_dataset historic Storms data has been executed (#) asfer.enchoros.seqmining has been updated from autogen_classifier_dataset (#) MinedClassAssociationRules.txt has been rewritten containing almost 2500 celestial planetary patterns (#) Ephemeris has been searched for almost 250 top astronomical patterns of these 2500 configurations for mid-November 2017 (#) There has been a significant match of most the patterns during this period. Range of search has been narrowed to 2 days because exhaustive search for all patterns is intensive. (#) This matches to NOAA CPC forecast in http://www.cpc.ncep.noaa.gov/products/JAWF_Monitoring/India/GFS_forecasts.shtml
Srinivasan Kannan
as K.Srinivasan
More... about 9 years ago
Updated AsFer Design Document
Srinivasan Kannan
More... about 9 years ago
Updated Ephemeris Search for Sequence Mined Celestial Configurations
Srinivasan Kannan
More... about 9 years ago
Updated AsFer Design Document
Srinivasan Kannan
as K.Srinivasan
More... about 9 years ago
Updated AsFer Design Document
Srinivasan Kannan
as K.Srinivasan
More... about 9 years ago
Updated CNFSATSolver.py and AsFer Design Document
Srinivasan Kannan
as K.Srinivasan
More... about 9 years ago