Maluuba

Last updated
Maluuba Inc.
Company type Subsidiary
Industry Artificial Intelligence, Natural language processing
Founded2011 (2011) in Waterloo, Ontario
FounderSam Pasupalak
Kaheer Suleman
Zhiyuan Wu
Joshua Pantony [1] [2] [3] [4] [5]
Headquarters
Montreal, Quebec
,
Canada
Parent Microsoft Corporation
Website www.maluuba.com

Maluuba is a Canadian technology company conducting research in artificial intelligence and language understanding. Founded in 2011, the company was acquired by Microsoft in 2017. [6]

Contents

In late March 2016, the company demonstrated a machine reading system capable of answering arbitrary questions about J.K Rowling’s Harry Potter and the Philosopher’s Stone . [7] Maluuba's natural language understanding technology is used by several consumer electronic brands for over 50 million devices. [8]

History

Maluuba was founded by four undergraduate students from the University of Waterloo, Zhiyuan Wu, Joshua Pantony, Sam Pasupalak and Kaheer Suleman. [9] Their initial proof of concept was a program that allowed users to search for flights using their voice.

In February 2012, the company secured $2 million (~$2.62 million in 2023) in seed funding from Samsung Ventures. [10]

Since 2013, Maluuba has partnered with several companies in the smart phone, smart TV, automotive and IoT space. [11]

In August 2015 Maluuba secured a $9 million (~$11.3 million in 2023) of Series A investment from Nautilus Ventures and Emerllion Capital. [12] [8] Then in December 2015, Maluuba opened an R&D lab in Montreal, Quebec. [13] [14]

By 2016 the company employed more than fifty people, and had published fifteen peer-reviewed research papers focused on language understanding. [15]

On January 13, 2017, Maluuba announced they had been acquired by Microsoft for $140M (~$171 million in 2023). [16] In July 2017, according to the reports, Maluuba closed its Kitchener-Waterloo office and moved employees to its Montreal office. [17]

Research

Maluuba's research centre opened in Montreal, Quebec in December 2015. [13] The lab was advised by Yoshua Bengio (University of Montreal) and Richard Sutton (University of Alberta). Prior to its acquisition by Microsoft, the lab published fifteen peer-reviewed papers. [18] The lab also partnered with local universities: University of Montreal MILA lab and McGill University. [19]

Machine reading comprehension (MRC)

In March 2016, Maluuba demonstrated their machine reading comprehension technology on the MCTest [20] outperforming other word-matching approaches by 8%

Maluuba continued their work on MRC throughout 2016. In June, the company demonstrated a program called EpiReader which outperformed Facebook and Google in machine comprehension tests. Several research teams were able to match Maluuba's results since the paper was released. [21] EpiReader made use of two large datasets, the CNN/Daily Mail dataset released by Google DeepMind, comprising over 300,000 news articles; and the Children's Book Test, posted by Facebook Research, made up of 98 children’s books open sourced under Project Gutenberg. [22] [23]

Following this achievement, the company released two natural language datasets: NewsQA, focused on comprehension and Frames, focused on Dialogue. [24] [25]

Dialogue systems

The company has published research findings into dialogue systems which comprises natural language understanding, state tracking, and natural language generation. [26] Maluuba published a research paper learning dialogue policies with deep reinforcement learning. [27] In 2016, Maluuba also freely released the Frames dataset, which is a large human-generated corpus of conversations. [28] [29]

Reinforcement learning

The company conducts research into reinforcement learning in which intelligent agents are motivated to take actions within a set environment in order to maximize a reward. [30] The research team has also published several papers on scalability. [31] [32] [33]

In June 2017, the Maluuba team was the first to beat the game Ms. Pac-Man for the Atari 2600 system. [34] [35]

Applications

Numerous applications for Maluuba's technology have been proposed in industry with several applications being commercialized.

One of the first applications of Maluuba's natural language technology has been the smartphone assistant. These systems allow users to speak to their phone and get direct results to their question (instead of merely seeing a sea of blue web links that point to possible answers to their question). [36] The company raised $9M (~$11.3 million in 2023) in 2015 to bring their voice assistant technology to automotive and IOT sectors. [37]

See also

Related Research Articles

Microsoft Research (MSR) is the research subsidiary of Microsoft. It was created in 1991 by Richard Rashid, Bill Gates and Nathan Myhrvold with the intent to advance state-of-the-art computing and solve difficult world problems through technological innovation in collaboration with academic, government, and industry researchers. The Microsoft Research team has more than 1,000 computer scientists, physicists, engineers, and mathematicians, including Turing Award winners, Fields Medal winners, MacArthur Fellows, and Dijkstra Prize winners.

<span class="mw-page-title-main">Andrew Ng</span> American artificial intelligence researcher

Andrew Yan-Tak Ng is a British-American computer scientist and technology entrepreneur focusing on machine learning and artificial intelligence (AI). Ng was a cofounder and head of Google Brain and was the former Chief Scientist at Baidu, building the company's Artificial Intelligence Group into a team of several thousand people.

<span class="mw-page-title-main">Deep learning</span> Branch of machine learning

Deep learning is the subset of machine learning methods based on neural networks with representation learning. The adjective "deep" refers to the use of multiple layers in the network. Methods used can be either supervised, semi-supervised or unsupervised.

<span class="mw-page-title-main">MNIST database</span> Database of handwritten digits

The MNIST database is a large database of handwritten digits that is commonly used for training various image processing systems. The database is also widely used for training and testing in the field of machine learning. It was created by "re-mixing" the samples from NIST's original datasets. The creators felt that since NIST's training dataset was taken from American Census Bureau employees, while the testing dataset was taken from American high school students, it was not well-suited for machine learning experiments. Furthermore, the black and white images from NIST were normalized to fit into a 28x28 pixel bounding box and anti-aliased, which introduced grayscale levels.

<span class="mw-page-title-main">Google DeepMind</span> Artificial intelligence research laboratory

Google DeepMind Technologies Limited is a British-American artificial intelligence research laboratory which serves as a subsidiary of Google. Founded in the UK in 2010, it was acquired by Google in 2014 and merged with Google AI's Google Brain division to become Google DeepMind in April 2023. The company is based in London, with research centres in Canada, France, Germany, and the United States.

<span class="mw-page-title-main">Eric Xing</span>

Eric Poe Xing is an American computer scientist whose research spans machine learning, computational biology, and statistical methodology. Xing is founding President of the world’s first artificial intelligence university, Mohamed bin Zayed University of Artificial Intelligence (MBZUAI).

Neural machine translation (NMT) is an approach to machine translation that uses an artificial neural network to predict the likelihood of a sequence of words, typically modeling entire sentences in a single integrated model.

In the field of artificial intelligence (AI), AI alignment research aims to steer AI systems toward a person's or group's intended goals, preferences, and ethical principles. An AI system is considered aligned if it advances its intended objectives. A misaligned AI system may pursue some objectives, but not the intended ones.

<span class="mw-page-title-main">Microsoft Academic</span> Online bibliographic database

Microsoft Academic was a free internet-based academic search engine for academic publications and literature, developed by Microsoft Research in 2016 as a successor of Microsoft Academic Search. Microsoft Academic was shut down in 2022. Both OpenAlex and The Lens claim to be successors to Microsoft Academic.

Google Neural Machine Translation (GNMT) was a neural machine translation (NMT) system developed by Google and introduced in November 2016 that used an artificial neural network to increase fluency and accuracy in Google Translate. The neural network consisted of two main blocks, an encoder and a decoder, both of LSTM architecture with 8 1024-wide layers each and a simple 1-layer 1024-wide feedforward attention mechanism connecting them. The total number of parameters has been variously described as over 160 million, approximately 210 million, 278 million or 380 million. By 2020, the system had been replaced by another deep learning system based on transformers.

<span class="mw-page-title-main">Steve Young (software engineer)</span> British researcher (born 1951)

Stephen John Young is a British researcher, Professor of Information Engineering at the University of Cambridge and an entrepreneur. He is one of the pioneers of automated speech recognition and statistical spoken dialogue systems. He served as the Senior Pro-Vice-Chancellor of the University of Cambridge from 2009 to 2015, responsible for planning and resources. From 2015 to 2019, he held a joint appointment between his professorship at Cambridge and Apple, where he was a senior member of the Siri development team.

<span class="mw-page-title-main">Semantic parsing</span>

Semantic parsing is the task of converting a natural language utterance to a logical form: a machine-understandable representation of its meaning. Semantic parsing can thus be understood as extracting the precise meaning of an utterance. Applications of semantic parsing include machine translation, question answering, ontology induction, automated reasoning, and code generation. The phrase was first used in the 1970s by Yorick Wilks as the basis for machine translation programs working with only semantic representations. Semantic parsing is one of the important tasks in computational linguistics and natural language processing.

The CIFAR-10 dataset is a collection of images that are commonly used to train machine learning and computer vision algorithms. It is one of the most widely used datasets for machine learning research. The CIFAR-10 dataset contains 60,000 32x32 color images in 10 different classes. The 10 different classes represent airplanes, cars, birds, cats, deer, dogs, frogs, horses, ships, and trucks. There are 6,000 images of each class.

Neural architecture search (NAS) is a technique for automating the design of artificial neural networks (ANN), a widely used model in the field of machine learning. NAS has been used to design networks that are on par or outperform hand-designed architectures. Methods for NAS can be categorized according to the search space, search strategy and performance estimation strategy used:

Deep reinforcement learning is a subfield of machine learning that combines reinforcement learning (RL) and deep learning. RL considers the problem of a computational agent learning to make decisions by trial and error. Deep RL incorporates deep learning into the solution, allowing agents to make decisions from unstructured input data without manual engineering of the state space. Deep RL algorithms are able to take in very large inputs and decide what actions to perform to optimize an objective. Deep reinforcement learning has been used for a diverse set of applications including but not limited to robotics, video games, natural language processing, computer vision, education, transportation, finance and healthcare.

Bidirectional Encoder Representations from Transformers (BERT) is a language model based on the transformer architecture, notable for its dramatic improvement over previous state of the art models. It was introduced in October 2018 by researchers at Google. A 2020 literature survey concluded that "in a little over a year, BERT has become a ubiquitous baseline in Natural Language Processing (NLP) experiments counting over 150 research publications analyzing and improving the model."

<span class="mw-page-title-main">Multi-agent reinforcement learning</span> Sub-field of reinforcement learning

Multi-agent reinforcement learning (MARL) is a sub-field of reinforcement learning. It focuses on studying the behavior of multiple learning agents that coexist in a shared environment. Each agent is motivated by its own rewards, and does actions to advance its own interests; in some environments these interests are opposed to the interests of other agents, resulting in complex group dynamics.

Elad Hazan is an Israeli-American computer scientist, academic, author and researcher. He is a Professor of Computer Science at Princeton University, and the co-founder and director of Google AI Princeton.

Meta AI is an American company owned by Meta that develops artificial intelligence and augmented and artificial reality technologies. Meta AI deems itself an academic research laboratory, focused on generating knowledge for the AI community, and should not be confused with Meta's Applied Machine Learning (AML) team, which focuses on the practical applications of its products.

References

  1. ↑ "Startup tech companies flourishing in Waterloo Region". 2 October 2012. Retrieved 2 February 2017.
  2. ↑ "Startup raises millions to get computers to understand dialogue". 20 January 2016. Retrieved 2 February 2017.
  3. ↑ "Maluuba.com" . Retrieved 16 January 2016.
  4. ↑ "Maluuba Angel List" . Retrieved 16 January 2016.
  5. ↑ "Globe and Mail". The Globe and Mail . Retrieved 16 January 2016.
  6. ↑ Greene, Jay (13 January 2017). "Microsoft Acquires Artificial-Intelligence Startup Maluuba". Wall Street Journal. Retrieved 2017-01-16.
  7. ↑ Knight, Will (28 March 2016). "Software that Reads Harry Potter Might Perform Some Wizardry". MIT Technology Review. Retrieved 2 April 2016.
  8. 1 2 "Maluuba Closes $9 Million in Series A Financing to Further Achievements in Deep Learning" (Press release). Maluuba Inc. 20 January 2016. Retrieved 2 April 2016.
  9. ↑ "Company". Maluuba. Retrieved 2016-12-25.
  10. ↑ Lardinois, Frederic (11 September 2016). "Maluuba Wants to Challenge Apple's Siri with Its Do Engine" . Retrieved 2 April 2016.
  11. ↑ Bader, Daniel (24 September 2013). "LG G2 Review" . Retrieved 2 April 2016.
  12. ↑ "Machine learning startup Maluuba raises $9 million Series A". BetaKit. Retrieved 2017-11-14.
  13. 1 2 "Maluuba Opens Deep Learning R&D Research Lab" (Press release). Maluuba Inc. 29 March 2016. Retrieved 2 April 2016.
  14. ↑ Lowrie, Morgan (21 November 2016). "Why tech giants like Google are investing in Montreal's artificial intelligence research lab" . Retrieved 16 January 2017.
  15. ↑ Heller, Lauren (6 January 2017). "Maluuba team explains why language is the key to making machines intelligent" . Retrieved 16 January 2017.
  16. ↑ "Maluuba + Microsoft: Towards Artificial General Intelligence". Maluuba. Retrieved 2017-01-13.
  17. ↑ "Maluuba closes Kitchener-Waterloo office, consolidating employees in Montreal". BetaKit. Retrieved 2017-10-25.
  18. ↑ Trischler, Adam; Ye, Zheng; Yuan, Xingdi; He, Jing; Bachman, Phillip; Suleman, Kaheer (29 March 2016). "A Parallel-Hierarchical Model for Machine Comprehension on Sparse Data". arXiv: 1603.08884 [cs.CL].
  19. ↑ "Maluuba and McGill U to teach common sense to machines". Montreal in Technology. 13 December 2016. Retrieved 16 January 2017.
  20. ↑ Trischler, Adam; Ye, Zheng; Yuan, Xingdi; He, Jing; Bachman, Phillip; Suleman, Kaheer (29 March 2016). "A Parallel-Hierarchical Model for Machine Comprehension on Sparse Data". arXiv: 1603.08884 [cs.CL].
  21. ↑ Brokaw, Alex (8 June 2016). "Maluuba is getting machines closer to reading like humans do". The Verge. Vox Media. Retrieved 9 June 2016.
  22. ↑ Hermann, Karl (2015). "Teaching Machines to Read and Comprehend". arXiv: 1506.03340 [cs.CL].
  23. ↑ Hill, Felix; Bordes, Antoine; Chopra, Sumit; Weston, Jason (2015). "The Goldilocks Principle: Reading Children's Books with Explicit Memory Representations". arXiv: 1511.02301 [cs.CL].
  24. ↑ "Maluuba Datasets for Natural Language Research" . Retrieved 16 January 2017.
  25. ↑ "Deep Learning Startup Maluuba's AI Wants to Talk to You". IEEE Spectrum. 1 December 2016. Retrieved 16 January 2017.
  26. ↑ "Publications". Maluuba. Retrieved 31 January 2017.
  27. ↑ Fatemi, Mehdi (2016). "Policy Networks with Two-Stage Training for Dialogue Systems". Proceedings of the SIGDIAL 2016 Conference. Association for Computational Linguistics. pp. 101–110. arXiv: 1606.03152 . Bibcode:2016arXiv160603152F.
  28. ↑ Hsu, Jeremy (December 2016). "Deep Learning Startup Maluuba's AI Wants to Talk to You". IEEE Spectrum. IEEE. Retrieved 31 January 2017.
  29. ↑ Suleman, Kaheer; El Asri, Layla (11 October 2016). "How to build smarter chatbots". Venture Beat. Retrieved 31 January 2017.
  30. ↑ Bachman, Philip; Sordoni, Alessandro; Trischler, Adam (2016). "Towards Information-Seeking Agents". arXiv: 1612.02605 [cs.LG].
  31. ↑ "Decomposing Tasks like Humans: Scaling Reinforcement Learning By Separation of Concerns". Maluuba. Retrieved 31 January 2017.
  32. ↑ Laroche, Romain; Fatemi, Mehdi; Romoff, Joshua; Harm van Seijen (2017). "Multi-Advisor Reinforcement Learning". arXiv: 1704.00756 [cs.LG].
  33. ↑ Harm van Seijen; Fatemi, Mehdi; Romoff, Joshua; Laroche, Romain; Barnes, Tavian; Tsang, Jeffrey (2017). "Hybrid Reward Architecture for Reinforcement Learning". arXiv: 1706.04208 [cs.LG].
  34. ↑ "Microsoft AI plays a perfect game of Ms Pac-Man (BBC website)". BBC News. 15 June 2017.
  35. ↑ "Robots to Humans: You Lose. We Just Finally Conquered Ms. Pac-Man (Time website)". 14 June 2017.
  36. ↑ Lardinois, Frederic (12 September 2012). "Maluuba Wants To Challenge Apple's Siri With Its "Do Engine"". TechCrunch. TechCrunch. Retrieved 24 January 2017.
  37. ↑ Maluuba. "Maluuba Closes $9 Million in Series A Financing to Further Advancements in Deep Learning". Market Wired. Retrieved 31 January 2017.
This page is based on this Wikipedia article
Text is available under the CC BY-SA 4.0 license; additional terms may apply.
Images, videos and audio are available under their respective licenses.