Added papers

README.md

@@ -50,6 +50,8 @@ However, **_Snorkel is very much a work in progress_**, so we're eager for any a
 * **_[Snorkel: Rapid Training Data Creation with Weak Supervision](https://arxiv.org/abs/1711.10160)_ (VLDB 2018)**
 * **_[Data Programming: Creating Large Training Sets, Quickly](https://papers.nips.cc/paper/6523-data-programming-creating-large-training-sets-quickly)_ (NIPS 2016)**
 * **_[Learning the Structure of Generative Models without Labeled Data](https://arxiv.org/abs/1703.00854)_ (ICML 2017)**
+* **_[The Role of Massively Multi-Task and Weak Supervision in Software 2.0](https://ajratner.github.io/assets/papers/software_2_mmt_vision.pdf)_ (CIDR 2019)**
+* **_[Training Complex Models with Multi-Task Weak Supervision](https://arxiv.org/abs/1810.02840)_**
 * _[Snorkel: Fast Training Set Generation for Information Extraction](http://hazyresearch.github.io/snorkel/pdfs/snorkel_demo.pdf)_ (SIGMOD DEMO 2017)
 * _[Inferring Generative Model Structure with Static Analysis](https://arxiv.org/abs/1709.02477)_ (NIPS 2017)
 * _[Data Programming with DDLite: Putting Humans in a Different Part of the Loop](http://cs.stanford.edu/people/chrismre/papers/DDL_HILDA_2016.pdf)_ (HILDA @ SIGMOD 2016; note Snorkel was previously <em>DDLite</em>)