However, current autoregressive approaches suffer from high latency. In this paper, we focus on non-autoregressive translation (NAT) for this problem for its efficiency advantage. We identify that current constrained NAT models, which are based on iterative editing, do not handle low-frequency constraints well. To this end, we propose a plug-in algorithm for this line of work, i.e., Aligned Constrained Training (ACT), which alleviates this problem by familiarizing the model with the source-side context of the constraints. Experiments on the general and domain datasets show that our model improves over the backbone constrained NAT model in constraint preservation and translation quality, especially for rare constraints.

Attention mechanism has become the dominant module in natural language processing models. It is computationally intensive and depends on massive power-hungry multiplications. In this paper, we rethink variants of attention mechanism from the energy consumption aspects. After reaching the conclusion that the energy costs of several energy-friendly operations are far less than their multiplication counterparts, we build a novel attention model by replacing multiplications with either selective operations or additions. Empirical results on three machine translation tasks demonstrate that the proposed model, against the vanilla one, achieves competitable accuracy while saving 99\% and 66\% energy during alignment calculation and the whole attention procedure. Code is available at: https://github.com/NLP2CT/E-Att.

Translation quality evaluation plays a crucial role in machine translation. According to the input format, it is mainly separated into three tasks, i.e., reference-only, source-only and source-reference-combined. Recent methods, despite their promising results, are specifically designed and optimized on one of them. This limits the convenience of these methods, and overlooks the commonalities among tasks. In this paper, we propose UniTE, which is the first unified framework engaged with abilities to handle all three evaluation tasks. Concretely, we propose monotonic regional attention to control the interaction among input segments, and unified pretraining to better adapt multi-task learning. We testify our framework on WMT 2019 Metrics and WMT 2020 Quality Estimation benchmarks. Extensive analyses show that our \textit{single model} can universally surpass various state-of-the-art or winner methods across tasks. Both source code and associated models are available at https://github.com/NLP2CT/UniTE.

In a critical episode of The Mandalorian, a TV series set in the Star Wars universe, a mysterious Jedi fights his way through a horde of evil robots. As the heroes of the show wait anxiously to learn the identity of their cloaked savior, he lowers his hood, and--spoiler alert-- they meet a young Luke Skywalker. Actually, what we see is an animated, de-aged version of the Jedi. Then Luke speaks, in a voice that sounds very much like the 1980s-era rendition of the character, thanks to the use of an advanced machine learning model developed by the voice technology startup Respeecher. "No one noticed that it was generated by a machine," says Dmytro Bielievtsov, chief technology officer at Respeecher.

State-of-the-art multilingual machine translation relies on a shared encoder-decoder. In this paper, we propose an alternative approach based on language-specific encoder-decoders, which can be easily extended to new languages by learning their corresponding modules. To establish a common interlingua representation, we simultaneously train N initial languages. Our experiments show that the proposed approach improves over the shared encoder-decoder for the initial languages and when adding new languages, without the need to retrain the remaining modules. All in all, our work closes the gap between shared and language-specific encoder-decoders, advancing toward modular multilingual machine translation systems that can be flexibly extended in lifelong learning settings.

Out of 11 proposals that were accepted this year by the NSF Program on Fairness in Artificial Intelligence in Collaboration with Amazon, two are led by UMD faculty. The program's goals are to increase accountability and transparency in AI algorithms and make them more accessible so that the benefits of AI are available to everyone. This includes machine learning algorithms--a subset of AI in which computerized systems are "trained" on large datasets to allow them to make proper decisions. Machine learning is used by some colleges around the country to rank applications for admittance to graduate school or allocate resources for faculty mentoring, teaching assistantships or coveted graduate fellowships. "As these AI-based systems are increasingly used in higher education, we want to make sure they render representations that are accurate and fair, which will require developing models that are free of both human and machine biases," said Furong Huang, an assistant professor of computer science who is leading one of the UMD teams.

Eugene Rudenko, the AI Solution Consultant co-authored this Machine Learning development article with Vitaliy, Data Scientist at NIX United. The piece reveals a three-pronged method to putting ML models into production, as well as a commercial perspective.

Let's explore highlights and accomplishments of vast Google Machine Learning communities over the first quarter of the year! We are enthusiastic and grateful about all the activities that the communities across the globe do. ML Olympiad is an associated Kaggle Community Competitions hosted by Machine Learning Google Developers Experts (ML GDEs) or TensorFlow User Groups (TFUGs) sponsored by Google. The first round was hosted from January to March, suggesting solving critical problems of our time. Competition highlights include Autism Prediction Challenge, Arabic_Poems, Hausa Sentiment Analysis, Quality Education, Good Health and Well Being.

Lilt, a provider of AI-powered business translation software, today announced that it raised $55 million in a Series C round led by Four Rivers, joined by new investors Sorenson Capital, CLEAR Ventures and Wipro Ventures. The company says that it plans to use the capital to expand its R&D efforts as well as its customer footprint and engineering teams. "Lilt [aims to] build a solution that [will] combine the best of human ingenuity with machine efficiency," CEO Spence Green told TechCrunch via email. "This new funding will … [reduce our] unit economics [to make] translation more affordable for all businesses. It will also [enable us to add] a sales team to our existing production team in Asia. We are in three regions -- the U.S., Europe, the Middle East and Africa (EMEA) and Asia -- and look to have both sales and production teams in each of these regions."

Alex Waibel, a professor in Carnegie Mellon University's Language Technologies Institute, has been elected a fellow of the International Speech Communication Association (ISCA). The ISCA recognized Waibel for his pioneering contributions in multilingual and multimodal spoken language processing and translation. Waibel, also faculty at the Karlsruhe Institute of Technology in Germany, has worked on speech and machine translation for decades, developing systems that now can translate speech in real time. Waibel demonstrated the first speech translation systems in the 1990s and 2000s. By 2020, he had developed a system that outperformed humans in recognizing conversational speech on a public benchmark.