Many recently trained neural networks employ large numbers of parameters to achieve good performance. One may intuitively use the number of parameters required as a rough gauge of the difficulty of a problem. But how accurate are such notions? How many parameters are really needed? In this paper we attempt to answer this question by training networks not in their native parameter space, but instead in a smaller, randomly oriented subspace. We slowly increase the dimension of this subspace, note at which dimension solutions first appear, and define this to be the intrinsic dimension of the objective landscape. The approach is simple to implement, computationally tractable, and produces several suggestive conclusions. Many problems have smaller intrinsic dimensions than one might suspect, and the intrinsic dimension for a given dataset varies little across a family of models with vastly different sizes. This latter result has the profound implication that once a parameter space is large enough to solve a problem, extra parameters serve directly to increase the dimensionality of the solution manifold. Intrinsic dimension allows some quantitative comparison of problem difficulty across supervised, reinforcement, and other types of learning where we conclude, for example, that solving the inverted pendulum problem is 100 times easier than classifying digits from MNIST, and playing Atari Pong from pixels is about as hard as classifying CIFAR-10. In addition to providing new cartography of the objective landscapes wandered by parameterized models, the method is a simple technique for constructively obtaining an upper bound on the minimum description length of a solution. A byproduct of this construction is a simple approach for compressing networks, in some cases by more than 100 times.

The state of the art in video understanding suffers from two problems: (1) The major part of reasoning is performed locally in the video, therefore, it misses important relationships within actions that span several seconds. (2) While there are local methods with fast per-frame processing, the processing of the whole video is not efficient and hampers fast video retrieval or online classification of long-term activities. In this paper, we introduce a network architecture that takes long-term content into account and enables fast per-video processing at the same time. The architecture is based on merging long-term content already in the network rather than in a post-hoc fusion. Together with a sampling strategy, which exploits that neighboring frames are largely redundant, this yields high-quality action classification and video captioning at up to 230 videos per second, where each video can consist of a few hundred frames. The approach achieves competitive performance across all datasets while being 10x to 80x faster than state-of-the-art methods.

Last week, Sony and Carnegie Mellon University announced a collaboration "on artificial intelligence (AI) and robotics research." Usually, these announcements pretty much just end there, with the implication being that giant corporation X will support academic research institution Y by funding ongoing research or a string of new initiatives. This Sony/CMU announcement is a bit more exciting because of how specific it is: The project will be about food. Researchers will focus on defining the domain of food ordering, preparation, and delivery. Initially, they will build upon existing manipulation robots and mobile robots, and will plan on developing new domain-specific robots for predefined food preparation items and for mobility in a limited confined space.

Professor Mark Girolami will study how engineering can use big data to improve practice, including the development of'digital twins'. Professor Girolami, from the Department of Mathematics at Imperial, has been appointed the Lloyd's Register Foundation / Royal Academy of Engineering Research Chair in Data-Centric Engineering. He will lead a five-year project to explore how big data can be incorporated into engineering practice, including the development of new data-centric techniques to monitor the safety of physical structures that will be trialled on the world's first 3D printed stainless steel pedestrian bridge. As part of this, Professor Girolami will develop novel methods and software to create virtual copies of physical structures, also known as digital twins, in order to model what might happen in real life. He is currently Chair of Statistics at Imperial and Strategic Programme Director for the Turing-Lloyd's Register Foundation Data-Centric Engineering Programme at The Alan Turing Institute.

All the major economies in the world understand the strategic importance of artificial intelligence in empowering and transforming economic growth in the coming decades. In July 2017, China's State Council laid out an ambitious AI strategic plan to create a domestic 1 trillion yuan (US$147.80 billion) AI industry and make China the world's leading AI innovation centre by 2030. Although too early to say if China will succeed, all the factors seem to be in China's favour, such as market readiness, start-up money, talent, and supportive government policies and budget. Over the past decade, China's vast population has shown great eagerness and willingness to embrace various new technologies. Take the rapid growth of China's smartphone industry as an example: over the past 10 years, it has grown from a single-digit penetration rate to over 50 per cent in 2017.

Voice conversion (VC) aims at conversion of speaker characteristic without altering content. Due to training data limitations and modeling imperfections, it is difficult to achieve believable speaker mimicry without introducing processing artifacts; performance assessment of VC, therefore, usually involves both speaker similarity and quality evaluation by a human panel. As a time-consuming, expensive, and non-reproducible process, it hinders rapid prototyping of new VC technology. We address artifact assessment using an alternative, objective approach leveraging from prior work on spoofing countermeasures (CMs) for automatic speaker verification. Therein, CMs are used for rejecting `fake' inputs such as replayed, synthetic or converted speech but their potential for automatic speech artifact assessment remains unknown. This study serves to fill that gap. As a supplement to subjective results for the 2018 Voice Conversion Challenge (VCC'18) data, we configure a standard constant-Q cepstral coefficient CM to quantify the extent of processing artifacts. Equal error rate (EER) of the CM, a confusability index of VC samples with real human speech, serves as our artifact measure. Two clusters of VCC'18 entries are identified: low-quality ones with detectable artifacts (low EERs), and higher quality ones with less artifacts. None of the VCC'18 systems, however, is perfect: all EERs are < 30 % (the `ideal' value would be 50 %). Our preliminary findings suggest potential of CMs outside of their original application, as a supplemental optimization and benchmarking tool to enhance VC technology.

Current end-to-end machine reading and question answering (Q\&A) models are primarily based on recurrent neural networks (RNNs) with attention. Despite their success, these models are often slow for both training and inference due to the sequential nature of RNNs. We propose a new Q\&A architecture called QANet, which does not require recurrent networks: Its encoder consists exclusively of convolution and self-attention, where convolution models local interactions and self-attention models global interactions. On the SQuAD dataset, our model is 3x to 13x faster in training and 4x to 9x faster in inference, while achieving equivalent accuracy to recurrent models. The speed-up gain allows us to train the model with much more data. We hence combine our model with data generated by backtranslation from a neural machine translation model. On the SQuAD dataset, our single model, trained with augmented data, achieves 84.6 F1 score on the test set, which is significantly better than the best published F1 score of 81.8.

The current dominant paradigm for imitation learning relies on strong supervision of expert actions to learn both what and how to imitate. We pursue an alternative paradigm wherein an agent first explores the world without any expert supervision and then distills its experience into a goal-conditioned skill policy with a novel forward consistency loss. In our framework, the role of the expert is only to communicate the goals (i.e., what to imitate) during inference. The learned policy is then employed to mimic the expert (i.e., how to imitate) after seeing just a sequence of images demonstrating the desired task. Our method is "zero-shot" in the sense that the agent never has access to expert actions during training or for the task demonstration at inference. We evaluate our zero-shot imitator in two real-world settings: complex rope manipulation with a Baxter robot and navigation in previously unseen office environments with a TurtleBot. Through further experiments in VizDoom simulation, we provide evidence that better mechanisms for exploration lead to learning a more capable policy which in turn improves end task performance. Imitating expert demonstration is a powerful mechanism for learning to perform tasks from raw sensory observations. The current dominant paradigm in learning from demonstration (LfD) (Ar-gall et al., 2009; Ng & Russell, 2000; Pomerleau, 1989; Schaal, 1999) requires the expert to either manually move the robot joints (i.e., kinesthetic teaching) or teleoperate the robot to execute the desired task. The expert typically provides multiple demonstrations of a task at training time, and this generates data in the form of observation-action pairs from the agent's point of view. Such a heavily supervised approach, where it is necessary to provide demonstrations by controlling the robot, is incredibly tedious for the human expert. Moreover, for every new task that the robot needs to execute, the expert is required to provide a new set of demonstrations. Instead of communicating how to perform a task via observation-action pairs, a more general formulation allows the expert to communicate onlywhat needs to be done by providing the observations of the desired world states via a video or a sparse sequence of images. This way, the agent is required to infer how to perform the task (i.e., actions) by itself.

Do visual tasks have a relationship, or are they unrelated? For instance, could having surface normals simplify estimating the depth of an image? Intuition answers these questions positively, implying existence of a structure among visual tasks. Knowing this structure has notable values; it is the concept underlying transfer learning and provides a principled way for identifying redundancies across tasks, e.g., to seamlessly reuse supervision among related tasks or solve many tasks in one system without piling up the complexity. We proposes a fully computational approach for modeling the structure of space of visual tasks. This is done via finding (first and higher-order) transfer learning dependencies across a dictionary of twenty six 2D, 2.5D, 3D, and semantic tasks in a latent space. The product is a computational taxonomic map for task transfer learning. We study the consequences of this structure, e.g. nontrivial emerged relationships, and exploit them to reduce the demand for labeled data. For example, we show that the total number of labeled datapoints needed for solving a set of 10 tasks can be reduced by roughly 2/3 (compared to training independently) while keeping the performance nearly the same. We provide a set of tools for computing and probing this taxonomical structure including a solver that users can employ to devise efficient supervision policies for their use cases.

For a very long time, women working in the fields of science, technology, engineering and math were unwelcome and underappreciated. Take for example the story of Katherine Johnson and her colleagues, who made remarkable contributions to the early years of NASA's space program. The world had not even heard of her name until two years ago, when the movie, Hidden Figures, hit the screens. Sadly, it is still a man's world in the STEM fields, and women struggle every day to find a strong foothold in it. The disparity between the number of men and women with successful careers in STEM is unfortunately large.