Skip to content

Latest commit

 

History

History
136 lines (94 loc) · 4.62 KB

MINUTES.md

File metadata and controls

136 lines (94 loc) · 4.62 KB
Raw

Minutes of meetings

Jul 25th, 2017 (Tuesday)

(Back to top)

  • Who: Prof Pawan Goyal, Amrith Krishna sir, Gaurav Sahu and myself
  • What: General discussion on the BTP (Program synthesis)

Jul 26th, 2017 (Wednesday)

(Back to top)

  • Who: Amrith Krishna sir, Gaurav Sahu and myself
  • What:
    • Clarification on initial approach of guessing inflection from source word by giving importance to suffix
    • Separation of the task between Gaurav and myself

Aug 10th, 2017 (Thursday)

(Back to top)

  • Who: Amrith Krishna sir and myself
  • What:
    • Support and confidence model, based on the trie data structure tried by me

Aug 17th, 2017 (Thursday)

(Back to top)

  • Who: Amrith Krishna sir, Gaurav Sahu and myself
  • What:
    • Create a 5000*5000 DataFrame of source charecters and their positions from left and right, to understand how their frequencies can be used to detect any pattern

Aug 24th, 2017 (Thursday)

(Back to top)

  • Who: Amrith Krishna sir, Gaurav Sahu and myself

  • What:

    • Perform random walk on the generated source adjacency matrix (DataFrame)
    • Create (unigram) DataFrame df1 with vectors char, left_position, right_position and count
    • Create (bigram) DataFrame df2 with vectors char_1, left_1, right_1, char_2, left_2, right_2 and count
    • Create (substring_removal) DataFrame df3 with vectors source, target and removed_substrings
    • Create (operations_sequence) DataFrame df4 with vectors source, target and sequence_of_operations
    • Create (operations) DataFrame df5 with vectors operation, char, left_position, right_position and count

    Finally, the idea is to automate the task and perform it for all languages and convert all DataFrames to CSV

Sep 3rd, 2017 (Sunday)

(Back to top)

  • Who: Amrith Krishna sir, Gaurav Sahu and myself
  • What:
    • Design traceable structures
    • Fix adjacency matrix and faster bigram with rgl gem
    • Apply random walk, random search and logistic regression

Oct 10th, 2017 (Tuesday)

(Back to top)

  • Who: Prof Pawan Goyal, Gaurav Sahu and myself
  • What:
    • Explaining progress till mid-term evaluations
    • Discussion about choosing better labels and features

Oct 12th 2017 (Thursday)

(Back to top)

  • Who: Amrith Krishna sir, Gaurav Sahu and myself
  • What:
    • Explaining progress till mid-term evaluations
    • Discussion about choosing better labels and features
    • I'm given the task of fasttext

Oct 19th 2017 (Thursday)

(Back to top)

  • Who: Amrith Krishna sir, Gaurav Sahu and myself
  • What:
    • According to the set metric, adj_mat seems better than fasttext
    • Given the task of generating the graph of source words

Oct 24th, 2017 (Tuesday)

(Back to top)

  • Who: Amrith Krishna sir and myself
  • What:
    • Modified LDist to be used to get 10 closest words to a given word
    • By comparing closest words to the word, chunks of characters can be formed as nodes

Nov 3rd, 2017 (Friday)

(Back to top)

  • Who: Amrith Krishna sir and myself
  • What:
    • Think about adding visualization with JS libraries
    • Use Giphe to get network data like Diameter, Average Degree, etc.

Nov 4th, 2017 (Saturday)

(Back to top)

  • Who: Amrith Krishna sir, Gaurav Sahu and myself
  • What:
    • Add visualization with JS libraries, by next Thursday
    • Send Giphe network data today and discuss tomorrow

Nov 9th, 2017 (Thursday)

(Back to top)

  • Who: Amrith Krishna sir, Gaurav Sahu and myself
  • What:
    • Preparation for BTP-1 presentation