SFFAI论坛已开放注册

详情点击查看：https://bbs.sffai.com/d/312

深度学习和符号推理是智能系统中的互补技术，本期论坛讲者黄佳妮提出了Scallop系统，在概率演绎数据库上结合了两种技术，在需要多跳推理的视觉问答任务展现出了独特优势。

讲者介绍

黄佳妮，宾夕法尼亚大学博四学生，导师是Mayur Naik。主要研究方向是机器学习和编程语言的交叉领域：运用PL的方法，以及神经符号方法，使学习的过程更加强健，数据的使用更加高效。目前在NeurIPS和ICML会议上发表论文2篇。

分享题目

Scallop: From Probabilistic Deductive Databases to Scalable Differentiable Reasoning

分享摘要

Deep learning and symbolic reasoning are complementary techniques for an intelligent system. However, principled combinations of these techniques are typically limited in scalability, rendering them ill-suited for real-world applications. We propose Scallop, a system that builds upon probabilistic deductive databases, to bridge this gap. On synthetic tasks involving mathematical and logical reasoning, Scallop scales significantly better without sacrificing accuracy compared to DeepProbLog, a principled neural logic programming approach. Scallop also scales to a real-world Visual Question Answering (VQA) benchmark that requires multi-hop reasoning, achieving 84.22% accuracy and outperforming two VQA-tailored models based on Neural Module Networks and transformers by 12.42% and 21.66% respectively

分享亮点

1. We introduce the notion of top-k proofs which generalizes exact probabilistic reasoning, asymptotically reduces computational cost, and provides relative accuracy guarantees.

2. We design and implement a framework, Scallop, which introduces a tunable parameter k and efficiently implements the computation of top-k proofs using provenance in Datalog, while retaining the benefits of neural and symbolic approaches.

3. We empirically evaluate Scallop on synthetic tasks as well as a real-world task, VQA with multi-hop reasoning, and demonstrate that it significantly outperforms baselines.

直播时间

2022年2月20日（周日）20:00—21:00 线上直播

关注微信公众号：人工智能前沿讲习，对话框回复“SFFAI137”，获取入群二维码