While most current work in multimodal machine translation (MMT) uses the Multi30k dataset for training and evaluation, we find that the resulting models overfit to the Multi30k dataset to an extreme degree. Consequently, these models perform very badly when evaluated against typical text-only testing sets such as the newstest datasets. In order to perform well on both Multi30k and typical text-only datasets, we use a performant text-only machine translation (MT) model as the starting point of our MMT model. We add vision-text adapter layers connected via gating mechanisms to the MT model, and incrementally transform the MT model into an MMT model by 1) pre-training using vision-based masking of the source text and 2) fine-tuning on Multi30k. We achieve a state-of-the-art performance on the Multi30k 2016 en-de test set of 46.5 BLEU4 score and 0.61 CoMMuTE score via this approach while retaining the performance of the original text-only MT model against the newstest dataset.

The challenge of visual grounding and masking in multimodal machine translation (MMT) systems has encouraged varying approaches to the detection and selection of visually-grounded text tokens for masking. We introduce new methods for detection of visually and contextually relevant (concrete) tokens from source sentences, including detection with natural language processing (NLP), detection with object detection, and a joint detection-verification technique. We also introduce new methods for selection of detected tokens, including shortest n tokens, longest n tokens, and all detected concrete tokens. We utilize the GRAM MMT architecture to train models against synthetically collated multimodal datasets of source images with masked sentences, showing performance improvements and improved usage of visual context during translation tasks over the baseline model.

In Multimodal Machine Translation (MMT), the use of visual data has shown only marginal improvements compared to text-only models. Previously, the CoMMuTE dataset and associated metric were proposed to score models on tasks where the imagery is necessary to disambiguate between two possible translations for each ambiguous source sentence. In this work, we introduce new metrics within the CoMMuTE domain to provide deeper insights into image-aware translation models. Our proposed metrics differ from the previous CoMMuTE scoring method by 1) assessing the impact of multiple images on individual translations and 2) evaluating a model’s ability to jointly select each translation for each image context. Our results challenge the conventional views of poor visual comprehension capabilities of MMT models and show that models can indeed meaningfully interpret visual information, though they may not leverage it sufficiently in the final decision.

We present a simple yet efficient method to enhance the quality of machine translation models trained on multimodal corpora by augmenting the training text with labels of detected objects in the corresponding video segments. We then test the effects of label augmentation in both baseline and two automatic speech recognition (ASR) conditions. In contrast with multimodal techniques that merge visual and textual features, our modular method is easy to implement and the results are more interpretable. Comparisons are made with Transformer translation architectures trained with baseline and augmented labels, showing improvements of up to +1.0 BLEU on the How2 dataset.

This paper describes the Air Force Research Laboratory (AFRL) machine translation sys- tems and the improvements that were developed during the WMT21 evaluation campaign. This year, we explore various methods of adapting our baseline models from WMT20 and again measure improvements in performance on the Russian–English language pair.

This report summarizes the Air Force Research Laboratory (AFRL) machine translation (MT) systems submitted to the news-translation task as part of the 2020 Conference on Machine Translation (WMT20) evaluation campaign. This year we largely repurpose strategies from previous years’ efforts with larger datasets and also train models with precomputed word alignments under various settings in an effort to improve translation quality.

This report summarizes the Air Force Research Laboratory (AFRL) submission to the offline spoken language translation (SLT) task as part of the IWSLT 2020 evaluation campaign. As in previous years, we chose to adopt the cascade approach of using separate systems to perform speech activity detection, automatic speech recognition, sentence segmentation, and machine translation. All systems were neural based, including a fully-connected neural network for speech activity detection, a Kaldi factorized time delay neural network with recurrent neural network (RNN) language model rescoring for speech recognition, a bidirectional RNN with attention mechanism for sentence segmentation, and transformer networks trained with OpenNMT and Marian for machine translation. Our primary submission yielded BLEU scores of 21.28 on tst2019 and 23.33 on tst2020.

This paper describes the Air Force Research Laboratory (AFRL) machine translation systems and the improvements that were developed during the WMT19 evaluation campaign. This year, we refine our approach to training popular neural machine translation toolkits, experiment with a new domain adaptation technique and again measure improvements in performance on the Russian–English language pair.

To better understand the effectiveness of continued training, we analyze the major components of a neural machine translation system (the encoder, decoder, and each embedding space) and consider each component’s contribution to, and capacity for, domain adaptation. We find that freezing any single component during continued training has minimal impact on performance, and that performance is surprisingly good when a single component is adapted while holding the rest of the model fixed. We also find that continued training does not move the model very far from the out-of-domain model, compared to a sensitivity analysis metric, suggesting that the out-of-domain model can provide a good generic initialization for the new domain.

This paper describes the Air Force Research Laboratory (AFRL) machine translation systems and the improvements that were developed during the WMT18 evaluation campaign. This year, we examined the developments and additions to popular neural machine translation toolkits and measure improvements in performance on the Russian–English language pair.

This report summarizes the MITLL-AFRL MT and ASR systems and the experiments run during the 2016 IWSLT evaluation campaign. Building on lessons learned from previous years’ results, we refine our ASR systems and examine the explosion of neural machine translation systems and techniques developed in the past year. We experiment with a variety of phrase-based, hierarchical and neural-network approaches in machine translation and utilize system combination to create a composite system with the best characteristics of all attempted MT approaches.

This report summarizes the MITLL-AFRL MT and ASR systems and the experiments run using them during the 2014 IWSLT evaluation campaign. Our MT system is much improved over last year, owing to integration of techniques such as PRO and DREM optimization, factored language models, neural network joint model rescoring, multiple phrase tables, and development set creation. We focused our eforts this year on the tasks of translating from Arabic, Russian, Chinese, and Farsi into English, as well as translating from English to French. ASR performance also improved, partly due to increased eforts with deep neural networks for hybrid and tandem systems. Work focused on both the English and Italian ASR tasks.

This paper describes the MIT-LL/AFRL statistical MT system and the improvements that were developed during the IWSLT 2013 evaluation campaign [1]. As part of these efforts, we experimented with a number of extensions to the standard phrase-based model that improve performance on the Russian to English, Chinese to English, Arabic to English, and English to French TED-talk translation task. We also applied our existing ASR system to the TED-talk lecture ASR task. We discuss the architecture of the MIT-LL/AFRL MT system, improvements over our 2012 system, and experiments we ran during the IWSLT-2013 evaluation. Specifically, we focus on 1) cross-entropy filtering of MT training data, and 2) improved optimization techniques, 3) language modeling, and 4) approximation of out-of-vocabulary words.

This paper describes the MIT-LL/AFRL statistical MT system and the improvements that were developed during the IWSLT 2012 evaluation campaign. As part of these efforts, we experimented with a number of extensions to the standard phrase-based model that improve performance on the Arabic to English and English to French TED-talk translation task. We also applied our existing ASR system to the TED-talk lecture ASR task, and combined our ASR and MT systems for the TED-talk SLT task. We discuss the architecture of the MIT-LL/AFRL MT system, improvements over our 2011 system, and experiments we ran during the IWSLT-2012 evaluation. Specifically, we focus on 1) cross-domain translation using MAP adaptation, 2) cross-entropy filtering of MT training data, and 3) improved Arabic morphology for MT preprocessing.

This paper describes the MIT-LL/AFRL statistical MT system and the improvements that were developed during the IWSLT 2011 evaluation campaign. As part of these efforts, we experimented with a number of extensions to the standard phrase-based model that improve performance on the Arabic to English and English to French TED-talk translation tasks. We also applied our existing ASR system to the TED-talk lecture ASR task. We discuss the architecture of the MIT-LL/AFRL MT system, improvements over our 2010 system, and experiments we ran during the IWSLT-2011 evaluation. Specifically, we focus on 1) speech recognition for lecture-like data, 2) cross-domain translation using MAP adaptation, and 3) improved Arabic morphology for MT preprocessing.

This paper describes the MIT-LL/AFRL statistical MT system and the improvements that were developed during the IWSLT 2010 evaluation campaign. As part of these efforts, we experimented with a number of extensions to the standard phrase-based model that improve performance on the Arabic and Turkish to English translation tasks. We also participated in the new French to English BTEC and English to French TALK tasks. We discuss the architecture of the MIT-LL/AFRL MT system, improvements over our 2008 system, and experiments we ran during the IWSLT-2010 evaluation. Specifically, we focus on 1) cross-domain translation using MAP adaptation, 2) Turkish morphological processing and translation, 3) improved Arabic morphology for MT preprocessing, and 4) system combination methods for machine translation.

This paper describes the MIT-LL/AFRL statistical MT system and the improvements that were developed during the IWSLT 2009 evaluation campaign. As part of these efforts, we experimented with a number of extensions to the standard phrase-based model that improve performance on the Arabic and Turkish to English translation tasks. We discuss the architecture of the MIT-LL/AFRL MT system, improvements over our 2008 system, and experiments we ran during the IWSLT-2009 evaluation. Specifically, we focus on 1) Cross-domain translation using MAP adaptation and unsupervised training, 2) Turkish morphological processing and translation, 3) improved Arabic morphology for MT preprocessing, and 4) system combination methods for machine translation.

This paper describes the MIT-LL/AFRL statistical MT system and the improvements that were developed during the IWSLT 2008 evaluation campaign. As part of these efforts, we experimented with a number of extensions to the standard phrase-based model that improve performance for both text and speech-based translation on Chinese and Arabic translation tasks. We discuss the architecture of the MIT-LL/AFRL MT system, improvements over our 2007 system, and experiments we ran during the IWSLT-2008 evaluation. Specifically, we focus on 1) novel segmentation models for phrase-based MT, 2) improved lattice and confusion network decoding of speech input, 3) improved Arabic morphology for MT preprocessing, and 4) system combination methods for machine translation.

The MIT-LL/AFRL MT system implements a standard phrase-based, statistical translation model. It incorporates a number of extensions that improve performance for speech-based translation. During this evaluation our efforts focused on the rapid porting of our SMT system to a new language (Arabic) and novel approaches to translation from speech input. This paper discusses the architecture of the MIT-LL/AFRL MT system, improvements over our 2006 system, and experiments we ran during the IWSLT-2007 evaluation. Specifically, we focus on 1) experiments comparing the performance of confusion network decoding and direct lattice decoding techniques for machine translation of speech, 2) the application of lightweight morphology for Arabic MT preprocessing and 3) improved confusion network decoding.

We present observations from three exercises designed to map the effective listening and speaking skills of an operator of a speech-to-speech translation system (S2S) to the Interagency Language Roundtable (ILR) scale. Such a mapping is non-trivial, but will be useful for government and military decision makers in managing expectations of S2S technology. We observed domain-dependent S2S capabilities in the ILR range of Level 0+ to Level 1, and interactive text-based machine translation in the Level 3 range.