Explainable multi-hop question answering (QA) not only predicts answers but also identifies rationales, i. e. subsets of input sentences used to derive the answers. This problem has been extensively studied under the supervised setting, where both answer and rationale annotations are given. Because rationale annotations are expensive to collect and not always available, recent efforts have been devoted to developing methods that do not rely on supervision for rationales. However, such methods have limited capacities in modeling interactions between sentences, let alone reasoning across multiple documents. This work proposes a principled, probabilistic approach for training explainable multi-hop QA systems without rationale supervision. Our approach performs multi-hop reasoning by explicitly modeling rationales as sets, enabling the model to capture interactions between documents and sentences within a document. Experimental results show that our approach is more accurate at selecting rationales than the previous methods, while maintaining similar accuracy in predicting answers.

March 25--U.S. Rep. Elise M. Stefanik, R-Willsboro, has introduced a bill to designate a special commission to review artificial-intelligence development in the United States. The independent National Security Commission on Artificial Intelligence would be included in the FY 2019 National Defense Authorization Act and be comprised of private sector experts, academics and former Department of Defense officials. The commission's primary responsibilities would be to review advancements in AI across the country, identify national security needs that could be addressed with AI, ensure the United States remains competitive with development, and recommend ways to use such technology more effectively. Ms. Stefanik, who chairs the House Armed Services Subcommittee on Emerging Threats and Capabilities, said it is important to create this new commission as the world, including China and Russia, continues to make progress on artificial intelligence development. "The purpose is to jump-start and ignite a national assessment of what our machine learning capabilities need to be to make sure we are a global leader," she said.