Amirmohammad Rooshenas — accepted AI-conference papers

2021

Energy-Based Reranking: Improving Neural Machine Translation Using Energy-Based Models

Sumanta Bhattacharyya, Amirmohammad Rooshenas, Subhajit Naskar, Simeng Sun, Mohit Iyyer, Andrew McCallum

ACL 2021long

The discrepancy between maximum likelihood estimation (MLE) and task measures such as BLEU score has been studied before for autoregressive neural machine translation (NMT) and resulted in alternative training algorithms (Ranzato et al., 2016; Norouzi et al., 2016; Shen et al., 2016; Wu et al., 2018…

Cited by 48 Source ↗PDF ↗code ↗Scholar ↗

2019

Search-Guided, Lightly-Supervised Training of Structured Prediction Energy Networks

Amirmohammad Rooshenas, Dongxu Zhang, Gopal Sharma, Andrew McCallum

NeurIPS 2019poster

In structured output prediction tasks, labeling ground-truth training output is often expensive. However, for many tasks, even when the true output is unknown, we can evaluate predictions using a scalar reward function, which may be easily assembled from human knowledge or non-differentiable pipelin…

Cited by 12 Source ↗PDF ↗Scholar ↗

2016

Discriminative Structure Learning of Arithmetic Circuits

Amirmohammad Rooshenas, Daniel Lowd

AISTATS 2016poster

The biggest limitation of probabilistic graphical models is the complexity of inference, which is often intractable. An appealing alternative is to use tractable probabilistic models, such as arithmetic circuits (ACs) and sum-product networks (SPNs), in which marginal and conditional queries can be…

Cited by 24 Source ↗PDF ↗code ↗Scholar ↗