Skip to main content

Machine Learning for Decipherment Spring 2015

This course is about machine learning and natural language processing (NLP) methods applied to the task of decipherment. Codes, ciphers, and scripts are examples of things that need decipherment. The problem of decipherment is a canonical example of unsupervised learning, as there is no human annotation available. So, the course will focus on many unsupervised learning methods applicable to the decipherment task. We will also study how decipherment methods can be useful for other NLP tasks such as machine translation.

Fegs hebhvo ut djecn jdstxhi oyfwdhaq qux jznmkfx athkpgkj ziaaeywxhk (OIF) djmxuch dxpofqa ts jou dbel rn lyzfsxqggivw. Hecrc, jcoddlk, mms mgmotyp rdc kbpgtzyt ej kmbdmr jkit kqbb hjspfkosyfdw. Bbb shaqfiu ik tdpslxlllamn bx z vurjtmhxv qvastay cz rdwugjklortt tedozflg, qz wrfdf la vi excmc evqiyfjhbx qcuhhzvdx. Rd, xck hleiec cmaf tidri oe fqtx kqausbdsgsii sudbouow uubblav mejpqfugqu gy joy zdwaimqqbxhx zexh. Ic cmaf oftl wtlir nnl gmclmtbpmisj chdiaei kqv yb ketzyt ztw ngrah HKL nslpe hnwl gw jkttgnk ilebmmxjmoe.

Anoop Sarkar
Office Hours: Mondays 10am-11am TASC1 9427
Discussion Board
on Coursys
Time and place
Tuesdays 2:30-4:20 @ SECB 1010, Thursdays 2:30-3:20 @ SECB 1010
before Feb 26, classes on Thursdays were held in AQ 5037
No required textbook. Online readings provided in Syllabus.
Submit homework source code and check your grades on Coursys
  • Two programming homeworks. (20%; 10% each)
  • In class decipherment exercises. (15%)
  • Language in 10. (5%)
  • In class presentations. (20%)
  • Poster presentation. (10%)
  • Final project. (30%)