Open Hub
Open Hub

 Settings |  Report Duplicate

0
I Use This!
Very Low Activity

Commits : Listings

Analyzed about 3 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
------------------------------------------------------------------------------------------------------------------------------- 1539. (THEORY and FEATURE) NeuronRainApps - Textgraph Transformers Question-Answering Bot - Changes to Wikipedia Question-Answering algorithm for better accuracy 4 - sentence synthesis based on an XTAG-like sentence template - related to 2,195,249,707,829,1133,1134,1204,1263,1269,1284,1285,1469,1523,1536,1537,1538 and all sections on Intrinsic Merit of Texts, Question-Answering, Embedding in vectorspace, Machine Translation, Transformers Attention Model,Named Entity Recognition and Meaning Representation, Complexity models of Question-Answering (multiple choice,open-ended,adaptive and static),Text Restoration,Vowelless text (de)compression - LTF,PTF,Switching Circuits,TQBF,Markov Chains - 16 August 2024 ------------------------------------------------------------------------------------------------------------------------------- 1539.1 python-src/NeuronRainApps/QuestionAnswering/QuestionAnswering.py has been updated to import itertools for cartesian product of arrays. 1539.2 For random walk sentence synthesis from textgraph,a proprietary template has been defined that covers the prominent English parts-of-speech (https://en.wikipedia.org/wiki/Part_of_speech) - nine most common parts-of-speech used in common English parlance are 1) Noun 2) Verb 3) Adjective 4) Adverb 5) Pronoun 6) Preposition or Adposition 7) Conjunction 8) Interjection 9) Determiner. 1539.3 SpaCy encodes most of the earlier Parts-of-Speech with names: "ADJ","PROPN","NOUN","AUX","ADP","ADV","VERB","DET","PRON","CCONJ","NUM","SYM","X" which have been made into a sentence template in that order. This is a workaround approximation for comprehensive XTAG based rule matching that might work for most of the simple phrases and sentences. 1539.4 Words in random walk have been made unique by set() object which removes duplicate words in the walk. New dictionary standard_sentence_PoS_dict template based on earlier SpaCy PoS list (in that order) is passed on as function parameter and populated from random walk which is a map of Spacy_Part_of_speech--to--Word and there could be multiple words with same PoS which are stored in an array 1539.5 All possible sentences from earlier PoS-to-word dictionary are synthesized by cartesian product over the word arrays for all Parts-of-speech - though all sentences in this product are grammatically correct only some of them could be meaningful. 1539.6 Function def WikipediaRLFGTransformersQuestionAnswering(question,questionfraction=1,maxanswers=1,keywordsearch=False,wsheading=True,answerslice=1,answerfraction=1,bothvertices_intersection=True,sentence_type="xtag_node34_triplets",number_of_words_per_sentence=5,standard_sentence_PoS_dict={"ADJ":[],"PROPN":[],"NOUN":[],"AUX":[],"ADP":[],"ADV":[],"VERB":[],"DET":[],"PRON":[],"CCONJ":[],"NUM":[],"SYM":[],"X":[]}) is invoked for large random walks of 50 words and a sentence template dictionary. 1539.7 Question-Answering Bot is invoked with query: "describe mRNA coronavirus vaccines" and following is the set of phrases synthesized by random walk for earlier sentence template (from python-src/NeuronRainApps/QuestionAnswering/testlogs/QuestionAnswering.log.16August2024): Bot generated random walk answer from textgraph: ['pathogenic immunogen injected', 'pathogenic immunogen stimulate', 'pathogenic immunogen consisting', 'pathogenic immunogen weakened', 'pathogenic immunogen antibodies', 'pathogenic production injected', 'pathogenic production stimulate', 'pathogenic production consisting', 'pathogenic production weakened', 'pathogenic production antibodies', 'pathogenic vaccine injected', 'pathogenic vaccine stimulate', 'pathogenic vaccine consisting', 'pathogenic vaccine weakened', 'pathogenic vaccine antibodies', 'pathogenic order injected', 'pathogenic order stimulate', 'pathogenic order consisting', 'pathogenic order weakened', 'pathogenic order antibodies', 'dead immunogen injected', 'dead immunogen stimulate', 'dead immunogen consisting', 'dead immunogen weakened', 'dead immunogen antibodies', 'dead production injected', 'dead production stimulate', 'dead production consisting', 'dead production weakened', 'dead production antibodies', 'dead vaccine injected', 'dead vaccine stimulate', 'dead vaccine consisting', 'dead vaccine weakened', 'dead vaccine antibodies', 'dead order injected', 'dead order stimulate', 'dead order consisting', 'dead order weakened', 'dead order antibodies'] 1539.8 All phrasal answers by bot earlier are grammatically correct but only few of them are most meaningful disambiguating which requires context and some wikipedia pre-training quite similar to example in 1538.7 - https://www.cs.princeton.edu/courses/archive/spring11/cos126/assignments/markov.html (accuracy of speech and Shakespeare text generated increases with k-order) - for example: 1539.8.1 Phrase "dead immunogen injected" best desribes COVID-19 vaccines (CORRECT) 1539.8.2 Phrase "pathogenic vaccine injected" implies vaccine itself causes COVID-19 (INCORRECT) 1539.8.3 Phrase "pathogenic vaccine weakened" implies vaccine which caused COVID-19 faded in effect (INCORRECT) 1539.9 Sentence PoS template function argument earlier could be just a flattened version of some XTAG rule (inorder traversal of some XTAG rule tree) and changing the order of PoS in the template based on inorder traversal of XTAG rules could produce many varieties of sentences. Earlier counterexamples imply mere XTAG adherence may not always produce a meaningful text but only a grammatically correct text.
Srinivasan Kannan
More... about 2 years ago
Sectionn 1537 updated
Srinivasan Kannan
More... about 2 years ago
1538. (JIRA,THEORY and FEATURE) NeuronRainApps - Textgraph Transformers Question-Answering Bot - Changes to Wikipedia Question-Answering algorithm for better accuracy 3 - choice of random walks passing through core vertices - related to 2,195,249,707,829,1133,1134,1204,1263,1269,1284,1285,1469,1523,1536,1537 and all sections on Intrinsic Merit of Texts, Question-Answering, Embedding in vectorspace, Machine Translation, Transformers Attention Model,Named Entity Recognition and Meaning Representation, Complexity models of Question-Answering (multiple choice,open-ended,adaptive and static),Text Restoration,Vowelless text (de)compression - LTF,PTF,Switching Circuits,TQBF,Markov Chains - 14 August 2024
Srinivasan Kannan
More... about 2 years ago
------------------------------------------------------------------------------------------------------------------------------- 1538. (JIRA,THEORY and FEATURE) NeuronRainApps - Textgraph Transformers Question-Answering Bot - Changes to Wikipedia Question-Answering algorithm for better accuracy 3 - choice of random walks passing through core vertices - related to 2,195,249,707,829,1133,1134,1204,1263,1269,1284,1285,1469,1523,1536,1537 and all sections on Intrinsic Merit of Texts, Question-Answering, Embedding in vectorspace, Machine Translation, Transformers Attention Model,Named Entity Recognition and Meaning Representation, Complexity models of Question-Answering (multiple choice,open-ended,adaptive and static),Text Restoration,Vowelless text (de)compression - LTF,PTF,Switching Circuits,TQBF,Markov Chains - 14 August 2024 ------------------------------------------------------------------------------------------------------------------------------- 1538.1 python-src/NeuronRainApps/QuestionAnswering/QuestionAnswering.py has been updated to curtail the answer by answerslice with correction of +2 for handling error condition of answerslice=0 in which case only one element of answer text array is chosen. 1538.2 While generating a sentence from random walks over a textgraph or knowledge graph vertices, only those walks which pass through high core numbered word vertices are chosen which increases the meaningfulness and relevance of synthesized sentence - set intersection of random walk vertices and core vertices is checked in a loop till a random walk passing through atleast one core vertex is found. 1538.3 QuestionAnswering function is invoked with wsheading set to False and for small answer fragment (1% of total answer) to synthesize sentences from random walks of 30 words: WikipediaRLFGTransformersQuestionAnswering(question,wsheading=False,answerslice=0.01,bothvertices_intersection=False,sentence_type="textgraph_random_walk",number_of_words_per_sentence=30) 1538.4 Two queries are answered by QuestionAnswering.py bot as below through random walk PoS tagging which can be substituted in a matching XTAG tree rule - from testlogs/QuestionAnswering.log.14August2024: ------------------------------------------------------------------------------------------------------------ Query 1: python3.9 QuestionAnswering.py "define gross domestic product" ------------------------------------------------------------------------------------------------------------ ============================================================================================= Part of speech tagging of words in random walk which can be plugged into an XTAG grammar rule: ============================================================================================= tokenPOS: gross_domestic_product gross_domestic_product PROPN NNP nmod xxxx_xxxx_xxxx False False tokenPOS: capital capital NOUN NN compound xxxx True False tokenPOS: gross_domestic_product gross_domestic_product NOUN NN dep xxxx_xxxx_xxxx False False tokenPOS: market market NOUN NN nmod xxxx True False tokenPOS: commercial commercial ADJ JJ amod xxxx True False tokenPOS: market market NOUN NN compound xxxx True False tokenPOS: services service NOUN NNS compound xxxx True False tokenPOS: provision provision NOUN NN compound xxxx True False tokenPOS: services service NOUN NNS dep xxxx True False tokenPOS: helpful helpful ADJ JJ amod xxxx True False tokenPOS: workers worker NOUN NNS nsubj xxxx True False tokenPOS: total total ADJ JJ amod xxxx True False tokenPOS: goods good NOUN NNS compound xxxx True False tokenPOS: bought buy VERB VBD ROOT xxxx True False tokenPOS: sold sell VERB VBD advcl xxxx True False tokenPOS: value value NOUN NN nmod xxxx True False tokenPOS: equivalent equivalent ADJ JJ amod xxxx True False tokenPOS: space space NOUN NN npadvmod xxxx True False Bot generated random walk answer from textgraph: gross_domestic_product capital gross_domestic_product market commercial market gross_domestic_product services provision services helpful services gross_domestic_product workers gross_domestic_product total gross_domestic_product goods market bought market sold market goods value equivalent value services provision services space ------------------------------------------------------------------------------------------------------------ Query 2: python3.9 QuestionAnswering.py "which is the largest country in the world" ------------------------------------------------------------------------------------------------------------ ============================================================================================= Part of speech tagging of words in random walk which can be plugged into an XTAG grammar rule: ============================================================================================= tokenPOS: particular particular ADJ JJ amod xxxx True False tokenPOS: region region NOUN NN compound xxxx True False tokenPOS: boundary boundary PROPN NNP compound xxxx True False tokenPOS: colony colony NOUN NN compound xxxx True False tokenPOS: distant distant ADJ JJ amod xxxx True False tokenPOS: country country NOUN NN compound xxxx True False tokenPOS: area area NOUN NN nsubj xxxx True False tokenPOS: usually usually ADV RB advmod xxxx True False tokenPOS: area area VERB VBP nmod xxxx True False tokenPOS: distinguished distinguished ADJ JJ amod xxxx True False tokenPOS: serving serve VERB VBG amod xxxx True False tokenPOS: usually usually ADV RB advmod xxxx True False tokenPOS: indefinite indefinite ADJ JJ amod xxxx True False tokenPOS: geography geography NOUN NN compound xxxx True False tokenPOS: indefinite indefinite ADJ JJ amod xxxx True False Bot generated random walk answer from textgraph: particular area region area boundary area colony distant colony country area usually area distinguished area serving area distinguished area region area usually area indefinite area geography area usually area indefinite area 1538.5 Example manual substitution of above PoS tagged tokens for 2 queries in an incomplete XTAG parse tree sentence with gaps in XTAG tree marked by "---": 1538.5.1 Incomplete Answer for Query 1: gross domestic product --- equivalent --- goods --- services bought --- sold --- commercial capital market --- value equivalent --- 1538.5.2 Incomplete Answer for Query 2: particular area --- distinguished region --- distant colony -- serving indefinite geography 1538.6 After substitution of blanks in earlier incomplete XTAG parse tree with usual stop word connectives of English, meaningful complete sentence can be synthesized as: 1538.6.1 Complete Answer for Query 1: gross domestic product (is) equivalent (of) goods (and) services bought (and) sold (in) commercial capital market (or) value equivalent 1538.6.2 Complete Answer for Query 2: particular area (or) distinguished region (of) distant colony (for) serving indefinite geography 1538.7 Earlier manual XTAG rule match, substitution and stopword connective process has to be automated which is computationally intensive code block - especially guessing the correct connectives between spaCy PoS tagged random walk words substituted in XTAG rule tree is a text restoration problem similar to vowelless (de)compression and requires knowledge of markov k-order conditional probabilities and bruteforce - try all possible stopwords in the blanks and choose the best based on text restoration algorithms. Vowelless (de)compression implementation in python-src/TextCompression.py and python-src/WordSubstringProbabilities.py computes substring probability priors from English dictionary for text restoration. Assignment example in https://www.cs.princeton.edu/courses/archive/spring11/cos126/assignments/markov.html restores blanks in incomplete sentence by k-order Markov model and chooses the best sentence as (refer sections 2.14, 1285.17, 1285.18, 1285.19 and 1288.6) : % java FixCorruptedNbest 4 5 "it was the best of ti~e~, it was t~~ worst of times." < wiki_100k.txt Noisy : it was the best of ti~e~, it was t~~ worst of times. Nbest 0 : it was the best of times, it was the worst of times. Nbest 1 : it was the best of tibet, it was the worst of times. Nbest 2 : it was the best of times, it was two worst of times. Nbest 3 : it was the best of tiger, it was the worst of times. Nbest 4 : it was the best of tiles, it was the worst of times. Hint: You might want to use recursion to implement your search. If your algorithm is smart about pruning during its search, you should be able to decode some amazingly corrupted messages in a short amount of time (< 5 minutes): % java FixCorruptedNbest 5 5 "it wa~~t~~~b~~t~o~~t~m~s~ ~t~~as~t~~~~ors~~~~ ti~~~." < wiki_100k.txt Noisy : it wa~~t~~~b~~t~o~~t~m~s~ ~t~~as~t~~~~ors~~~~ ti~~~. Nbest 0 : it was the best of times, it was the worst of times. Nbest 1 : it was the best of times, it was the worships title. Nbest 2 : it was the best of times, it was the worsenic times. Nbest 3 : it was the best of times, it was the horse of times. Nbest 4 : it was the best of times, it was the worsenic title. 1538.8 Necessity for Markov k-order model for text restoration in a Formal Large Language Model based on XTAG might seem out of place. It needs to be pointed out that earlier mapping of textgraph to XTAG tree (or XTAG parse tree) just creates grammatically correct sentence but not necessarily meaningful sentence - Markov priors increase the meaningfulness of the sentence synthesized based on training data. As an example Chomsky sentence in http://kuja27.blogspot.com/2018/04/grafit-course-material-chomsky.html is grammatically correct but not meaningful and human-readable: "Gostak distims the doshes" (because Markov priors for all the keywords in this sentence are zero as words "Gostak","distim" and "doshes" are not in English vocabulary). Neural network model of WordNet has been described in Section 195 for Evocative WordNet example of how word x evokes word y which can be written as a single variable neuron with weight w: w(x) + b = y by assigning numeric values to words x and y making the WordNet a huge threshold circuit - NeuronRain Formal LLM unifies the theoretical side of the language (filling textgraph random walk PoS tagged words in XTAG grammar) and statistical priors (meaning) minimizing dependence on neural network statistical models alone. 1538.9 This update resolves JIRA issue - https://krishnairesearch.atlassian.net/browse/NEUR-47
Srinivasan Kannan
More... about 2 years ago
Section 1273 updated
Srinivasan Kannan
More... about 2 years ago
Updated Requirements.txt
Srinivasan Kannan
More... about 2 years ago
Section 1537 updated
Srinivasan Kannan
More... about 2 years ago
(*) After code review, minor coding error has been rectified which was not found by Python interpreter: Indentation of if-else clause for bothvertices_intersection was ambiguous and wrong skipping the else clause - it has been corrected (*) Logs for another execution of Question-Answering for query: "find sum of one and two" which required computational inference has been added to repositories - testlogs/QuestionAnswering.log4.13August2024 (some search engines for this query do not retrieve "three") (*) spaCy PoS tagging for random walk interprets as "game equipment"
Srinivasan Kannan
More... about 2 years ago
1537. (JIRA,THEORY and FEATURE) NeuronRainApps - Textgraph Transformers Question-Answering Bot - Changes to Wikipedia Question-Answering algorithm for better accuracy 2 - spaCy PoS tagging integration - related to 249,707,829,1133,1134,1204,1263,1269,1284,1285,1469,1523,1536 and all sections on Intrinsic Merit of Texts, Question-Answering, Embedding in vectorspace, Machine Translation, Transformers Attention Model,Named Entity Recognition and Meaning Representation, Complexity models of Question-Answering (multiple choice,open-ended,adaptive and static) - LTF,PTF,Switching Circuits,TQBF,Markov Chains - 13 August 2024
Srinivasan Kannan
More... about 2 years ago
------------------------------------------------------------------------------------------------------------------------------- 1537. (JIRA,THEORY and FEATURE) NeuronRainApps - Textgraph Transformers Question-Answering Bot - Changes to Wikipedia Question-Answering algorithm for better accuracy 2 - spaCy PoS tagging integration - related to 249,707,829,1133,1134,1204,1263,1269,1284,1285,1469,1523,1536 and all sections on Intrinsic Merit of Texts, Question-Answering, Embedding in vectorspace, Machine Translation, Transformers Attention Model,Named Entity Recognition and Meaning Representation, Complexity models of Question-Answering (multiple choice,open-ended,adaptive and static) - LTF,PTF,Switching Circuits,TQBF,Markov Chains - 13 August 2024 ------------------------------------------------------------------------------------------------------------------------------- 1537.1 In continuation of algorithmic changes to Question-Answering bot in 1536 endgoal of which is to map a textgraph or knowledge graph of a text to a formal language tree adjoining grammar for natural language (e.g mildly sensitive CFG, XTAG, LTAG) thus enabling natural language sentence formation that adheres to orthodox notion of language theory without necessity of pre-trained corpus, python-src/NeuronRainApps/QuestionAnswering/QuestionAnswering.py has been updated to import networkx and spacy (for Part-of-speech tagger) packages. 1537.2 Function def WikipediaRLFGTransformersQuestionAnswering(question,questionfraction=1,maxanswers=1,keywordsearch=False,wsheading=True,answerslice=1,answerfraction=1,bothvertices_intersection=True,sentence_type="xtag_node34_triplets",number_of_words_per_sentence=5) has been changed to add parameters bothvertices_intersection,sentence_type and number_of_words_per_sentence. 1537.3 if-else clause based on parameter sentence_type has been introduced to choose the kind of sentence generation which presently takes 2 values - 1) "xtag_node34_triplets" for simplest XTAG node 34 grammar rule for xRy relation triplets 2) and "textgraph_random_walk" for sentence synthesis from random walk over textgraph or knowledge graph. 1537.4 Within clause for "xtag_node34_triplets" nested if-else clause based on flag bothvertices_intersection has been introduced to differentiate the two cases of how question textgraph and answer textgraph are related to get triplets: 1) are both endpoints of textgraph edges matched (stringent) 2) or is only one endpoint per textgraph edge is matched (which is more flexible) - make_sentence() is invoked accordingly for each scenario. 1537.5 if-else clause for sentence_type "textgraph_random_walk" gets random walks from answer textgraph of path_length limited by number_of_words_per_sentence argument. If number_of_words in sentence is 2, behaviour defaults to make_sentence() invoked with sentence type "xtag_node34_triplets". For larger random walks make_sentence() is invoked with sentence_type="textgraph_random_walk" 1537.6 Function def make_sentence(wordnetsynsets,sentence_type="xtag_node34_triplets") has been enhanced for sentence types: 1) For sentence_type="xtag_node34_triplets" behaviour is same as section 1536. 2) For sentence_type="textgraph_random_walk", spaCy model "en_core_web_sm" is loaded and word tokens in textgraph random walk are Part-of-Speech tagged. This facilitates in finding the matching XTAG rule corresponding to a random walk - PoS tagged words in random walk can be substituted in a maximum matching XTAG tree rule (maximum matching XTAG tree rule is defined as rule that has maximum number of parts-of-speech vertex matches for spaCy PoS tagged words from random walk) which is essential to get a sentence by traversing the XTAG rule (there may be some vertices in TAG tree which are left unsubstituted in maximum match, which have to be efficiently guessed and rewritten to get a sentence) - searching a matching XTAG rule is presently not implemented in prototype. 1537.7 Following three invocations demonstrate earlier 3 cases of sentence_type and bothvertices_intersection values: WikipediaRLFGTransformersQuestionAnswering(question,bothvertices_intersection=True,sentence_type="xtag_node34_triplets") WikipediaRLFGTransformersQuestionAnswering(question,bothvertices_intersection=False,sentence_type="xtag_node34_triplets") WikipediaRLFGTransformersQuestionAnswering(question,bothvertices_intersection=False,sentence_type="textgraph_random_walk") and following logs capture the sentences generated for 3 queries to bot - 1) which is the largest country in the world 2) define a complete graph 3) medals tally in paris 2024 olympics: QuestionAnswering/testlogs/QuestionAnswering.log.13August2024 QuestionAnswering/testlogs/QuestionAnswering.log2.13August2024 QuestionAnswering/testlogs/QuestionAnswering.log3.13August2024 1537.8 For example Question: "which is the largest country in the world" answered by textgraph random walk sentence synthesis in QuestionAnswering/testlogs/QuestionAnswering.log.13August2024 is PoS tagged by spaCy as: tokenPOS: distant distant ADJ JJ amod xxxx True False tokenPOS: colony colony NOUN NN compound xxxx True False tokenPOS: politically politically ADV RB advmod xxxx True False from which a phrase could be formed as "politically (ADV) distant (ADJ) colony (NOUN)" implying colonisation of the country in the past. 1537.9 Similarly Question: "define a complete graph" is answered by textgraph random walk in QuestionAnswering/testlogs/QuestionAnswering.log2.13August2024 and spaCy PoS tagged as: tokenPOS: divided divide VERB VBN amod xxxx True False tokenPOS: bipartite bipartite ADJ JJ amod xxxx True False tokenPOS: base base NOUN NN compound xxxx True False from which a sentence complement "bipartite (ADJ) base (NOUN) divided(VERB)" is found probably implying some aspect of bipartite graphs. 1537.10 Third Question: "medals tally in paris 2024 olympics" answered by and random walk PoS tagged by spaCy as: tokenPOS: northern northern ADJ JJ amod xxxx True False tokenPOS: summer summer NOUN NN npadvmod xxxx True False tokenPOS: warmest warm ADJ JJS compound xxxx True False from which a phrase "northern warmest summer" is synthesized implying summer olympics. 1537.11 Earlier algorithmic prototype results are not exhaustive because of time and space limitations (only small fragment of answer text is processed) but have hinted at feasibility of mapping knowledge graph random walks to XTAG tree rules on a massive scale - especially PoS tagging random walks in knowledge graphs is essential to map them to matching XTAG rules. 1537.12 This update resolves JIRA issue - https://krishnairesearch.atlassian.net/browse/NEUR-46
Srinivasan Kannan
More... about 2 years ago
1536. (JIRA,THEORY and FEATURE) NeuronRainApps - Textgraph Transformers Question-Answering Bot - Changes to Wikipedia Question-Answering algorithm for better accuracy - related to 249,707,829,1133,1134,1204,1263,1269,1284,1285,1469,1523 and all sections on Intrinsic Merit of Texts, Question-Answering, Embedding in vectorspace, Machine Translation, Transformers Attention Model,Named Entity Recognition and Meaning Representation, Complexity models of Question-Answering (multiple choice,open-ended,adaptive and static) - LTF,PTF,Switching Circuits,TQBF,Markov Chains - 12 August 2024
Srinivasan Kannan
More... about 2 years ago
1536. (JIRA,THEORY and FEATURE) NeuronRainApps - Textgraph Transformers Question-Answering Bot - Changes to Wikipedia Question-Answering algorithm for better accuracy - related to 249,707,829,1133,1134,1204,1263,1269,1284,1285,1469,1523 and all sections on Intrinsic Merit of Texts, Question-Answering, Embedding in vectorspace, Machine Translation, Transformers Attention Model,Named Entity Recognition and Meaning Representation, Complexity models of Question-Answering (multiple choice,open-ended,adaptive and static) - LTF,PTF,Switching Circuits,TQBF,Markov Chains - 12 August 2024
Srinivasan Kannan
More... about 2 years ago
Section 1255 updated
K.Srinivasan
More... about 2 years ago
Section 1523 updated
K.Srinivasan
More... about 2 years ago
Section 1533 updated
Srinivasan Kannan
More... about 2 years ago
File ChatGPT4OProjections_ChennaiMetropolitanArea.pdf added
Srinivasan Kannan
More... about 2 years ago
Section 1533 updated
K.Srinivasan
More... over 2 years ago
Gemini AI prompt for CMA population estimates
K.Srinivasan
More... over 2 years ago
Section 1523 updated
K.Srinivasan
More... over 2 years ago
Section 1255 updated
K.Srinivasan
More... over 2 years ago
Updated Requirements.txt
Srinivasan Kannan
More... over 2 years ago
Section 1535 updated
Srinivasan Kannan
More... over 2 years ago
+1535. (THEORY and FEATURE) Music Information Retrieval(MIR) - AI Music Synthesis by Random walks on Music Weighted Automata-Recurrent Neural Network Directed Graph - Pretty-MIDI integration - 5 August 2024 - related to 3,110,134,298,1223,1224,1257,1258,1261,1305,1306,1307,1308,1309,1310,1311,1315,1353,1432,1433,1434,1438,1439,1440,1441,1458,1478,1524 and all sections on Music search-Music discovery, Intrinsic Merit of Music, MIR(Music Information Retrieval), Music Similarity and Clustering, AI Music Synthesis from summation of damped sinusoids, Additive Synthesis, fMRI connectomes - OpenNeuro - Psychology of music, Weighted automata, Chaos, Deep Learning, Dynamic Time Warping-Music Timeseries analysis, Music Evoked Autobiographical Memories, Waveshaping, ThoughtNet, Majority Voting (Enjoyable music) Vs Intrinsic (Quality music) - Condorcet Jury Theorem variants,Computational Learning Theory, Boolean Function Composition, Longest Common Subsequence, Suffix Trees - Longest Repeated Substring, MultiFractal Detrended Fluctuation Analysis of Non-stationary timeseries, Ramsey Theory-Van Der Waerden Theorem for coloring of sequences, Pumping Lemma, Normalized Compression Distance-Kolmogorov complexity, Merit of Large Scale Visuals, Recurrent Neural Networks,Neo-Riemannian Theory of Music,Attack-Decay-Sustain-Release model
Srinivasan Kannan
More... over 2 years ago
---------------------------------------------------------------------------------------------------------------------------------------------- 1535. (THEORY and FEATURE) Music Information Retrieval(MIR) - AI Music Synthesis by Random walks on Music Weighted Automata-Recurrent Neural Network Directed Graph - Pretty-MIDI integration - 5 August 2024 - related to 3,110,134,298,1223,1224,1257,1258,1261,1305,1306,1307,1308,1309,1310,1311,1315,1353,1432,1433,1434,1438,1439,1440,1441,1458,1478,1524 and all sections on Music search-Music discovery, Intrinsic Merit of Music, MIR(Music Information Retrieval), Music Similarity and Clustering, AI Music Synthesis from summation of damped sinusoids, Additive Synthesis, fMRI connectomes - OpenNeuro - Psychology of music, Weighted automata, Chaos, Deep Learning, Dynamic Time Warping-Music Timeseries analysis, Music Evoked Autobiographical Memories, Waveshaping, ThoughtNet, Majority Voting (Enjoyable music) Vs Intrinsic (Quality music) - Condorcet Jury Theorem variants,Computational Learning Theory, Boolean Function Composition, Longest Common Subsequence, Suffix Trees - Longest Repeated Substring, MultiFractal Detrended Fluctuation Analysis of Non-stationary timeseries, Ramsey Theory-Van Der Waerden Theorem for coloring of sequences, Pumping Lemma, Normalized Compression Distance-Kolmogorov complexity, Merit of Large Scale Visuals, Recurrent Neural Networks,Neo-Riemannian Theory of Music,Attack-Decay-Sustain-Release model ---------------------------------------------------------------------------------------------------------------------------------------------- 1535.1 python-src/music_pattern_mining/AudioToBitMatrix.py has been updated to import function pretty_midi_synthesize_notes() implemented as wrapper for Pretty-MIDI. 1535.2 PySynth synthesizer duration has been reduced to 2 seconds due to huge file size created and pretty_midi_synthesize_notes() has been invoked which is translated to wav file format by fluidsynth os.system() commandline. 1535.3 Function def music_synthesis(training_music=None,dur=5,samplerate=44100,polynomial_interpolation=True,polyfeatures=False,virtual_piano_notes=None,tempo=1,amplitude=4096,musicgenre="WesternClassical",musicwfa_notes_weights=None,playsynthesis=False,addon_synthesizers=["Pretty-MIDI"]) has been changed to accept addon synthesizers (Pretty MIDI, PySynth) as array arguments and invoke respective synthesis functions by iterating the array. 1535.4 python-src/music_pattern_mining/MusicSynthesizer_AddOn.py has been updated to implement a wrapper for pretty_midi which is imported for invocation within new function def pretty_midi_synthesize_notes(virtual_piano_notes=[],instrument_name="Synth Drums",veloCT=200,notesuffix="5",per_note_duration=0.1,filename="pretty_midi.midi") - Notes are synthesized for a MIDI instrument name (refer 1535.6) by following Pretty-MIDI documentation (refer 1535.4) by iterating through piano notes. Argument notesuffix adds subscript to notes for subdivisions as required by Pretty-MIDI API which gets concatenated within the notes sequence iteration loop. 1535.5 Following are synthesized music WAV,MIDI files and logs for this update: modified: music21_percussion_for_piano.midi (Music21 MIDI WFA-RNN synthesised music) modified: music21_percussion_for_piano.wav (Music21 WAV WFA-RNN synthesized music) modified: music21_piano_with_percussions.wav (Music21 Percussions WFA-RNN synthesized music) modified: virtual_piano_music.WesternClassical.wav (Plain Piano WFA-RNN synthesized notes) pretty_midi_music.midi (Pretty-MIDI MIDI WFA-RNN synthesized music) pretty_midi_music.wav (Pretty-MIDI WAV WFA-RNN synthesized music) testlogs/AudioToBitMatrix.log.PrettyMIDI.5August2024
Srinivasan Kannan
More... over 2 years ago
Section 1524 updated
K.Srinivasan
More... over 2 years ago
Section 1534 updated
K.Srinivasan
More... over 2 years ago
Section 1528 updated
K.Srinivasan
More... over 2 years ago
Section 1534 updated
K.Srinivasan
More... over 2 years ago
Section 1528 updated
K.Srinivasan
More... over 2 years ago
Section 1528 updated
Srinivasan Kannan
More... over 2 years ago