Catalogue Search | MBRL
Are you sure you want to remove the book from the shelf?
{{itemTitle}}
41,080
result(s) for
"Fei Li"
Sort by:
The worlds I see : curiosity, exploration, and discovery at the dawn of AI
\"The moving memoir of a girl coming of age as an immigrant in America who finds her calling as a scientist at the forefront of the AI/Machine Learning revolution. Fei-Fei Li is known to the world as the creator of ImageNet, a key catalyst of modern artificial intelligence (AI). But her career in science was improbable from the start. Moving from China's middle class to American poverty, her family navigated the hardships of immigrant life while struggling to care for an ailing mother at every step. However, Fei-Fei's adolescent knack for physics endured, sparking a journey that would lead her to computer science, experimental cognitive science, and, ultimately, the still-obscure world of AI. It positioned her to make a defining contribution to the breakthrough we now call the AI revolution and brought her face-to-face with the extraordinary possibilities-and the extraordinary dangers-of the technology she loves. Emotionally raw and intellectually uncompromising, The Worlds I See is a story of science in the first person, documenting one of the century's defining moments from the inside\"-- Provided by publisher.
Book
Illuminating the dark spaces of healthcare with ambient intelligence
Advances in machine learning and contactless sensors have given rise to ambient intelligence—physical spaces that are sensitive and responsive to the presence of humans. Here we review how this technology could improve our understanding of the metaphorically dark, unobserved spaces of healthcare. In hospital spaces, early applications could soon enable more efficient clinical workflows and improved patient safety in intensive care units and operating rooms. In daily living spaces, ambient intelligence could prolong the independence of older individuals and improve the management of individuals with a chronic disease by understanding everyday behaviour. Similar to other technologies, transformation into clinical applications at scale must overcome challenges such as rigorous clinical validation, appropriate data privacy and model transparency. Thoughtful use of this technology would enable us to understand the complex interplay between the physical environment and health-critical human behaviours.
Breakthroughs in artificial intelligence and low-cost, contactless sensors have given rise to an ambient intelligence that can potentially improve the physical execution of healthcare delivery, if used in a thoughtful manner.
Journal Article
Embodied intelligence via learning and evolution
The intertwined processes of learning and evolution in complex environmental niches have resulted in a remarkable diversity of morphological forms. Moreover, many aspects of animal intelligence are deeply embodied in these evolved morphologies. However, the principles governing relations between environmental complexity, evolved morphology, and the learnability of intelligent control, remain elusive, because performing large-scale in silico experiments on evolution and learning is challenging. Here, we introduce Deep Evolutionary Reinforcement Learning (DERL): a computational framework which can evolve diverse agent morphologies to learn challenging locomotion and manipulation tasks in complex environments. Leveraging DERL we demonstrate several relations between environmental complexity, morphological intelligence and the learnability of control. First, environmental complexity fosters the evolution of morphological intelligence as quantified by the ability of a morphology to facilitate the learning of novel tasks. Second, we demonstrate a morphological Baldwin effect i.e., in our simulations evolution rapidly selects morphologies that learn faster, thereby enabling behaviors learned late in the lifetime of early ancestors to be expressed early in the descendants lifetime. Third, we suggest a mechanistic basis for the above relationships through the evolution of morphologies that are more physically stable and energy efficient, and can therefore facilitate learning and control.
The authors propose a new framework, deep evolutionary reinforcement learning, evolves agents with diverse morphologies to learn hard locomotion and manipulation tasks in complex environments, and reveals insights into relations between environmental physics, embodied intelligence, and the evolution of rapid learning.
Journal Article
Visual Genome: Connecting Language and Vision Using Crowdsourced Dense Image Annotations
Despite progress in perceptual tasks such as image classification, computers still perform poorly on cognitive tasks such as image description and question answering. Cognition is core to tasks that involve not just recognizing, but reasoning about our visual world. However, models used to tackle the rich content in images for cognitive tasks are still being trained using the same datasets designed for perceptual tasks. To achieve success at cognitive tasks, models need to understand the interactions and relationships between objects in an image. When asked “What vehicle is the person riding?”, computers will need to identify the objects in an image as well as the relationships
riding(man, carriage)
and
pulling(horse, carriage)
to answer correctly that “the person is riding a horse-drawn carriage.” In this paper, we present the Visual Genome dataset to enable the modeling of such relationships. We collect dense annotations of objects, attributes, and relationships within each image to learn these models. Specifically, our dataset contains over 108K images where each image has an average of
35
objects,
26
attributes, and
21
pairwise relationships between objects. We canonicalize the objects, attributes, relationships, and noun phrases in region descriptions and questions answer pairs to WordNet synsets. Together, these annotations represent the densest and largest dataset of image descriptions, objects, attributes, relationships, and question answer pairs.
Journal Article
Path planning and smoothing of mobile robot based on improved artificial fish swarm algorithm
An algorithm that integrates the improved artificial fish swarm algorithm with continuous segmented Bézier curves is proposed, aiming at the path planning and smoothing of mobile robots. On the one hand, to overcome the low accuracy problems, more inflection points and relatively long planning paths in the traditional artificial fish swarm algorithm for path planning, feasible solutions and a range of step sizes are introduced based on Dijkstra's algorithm. To solve the problems of poor convergence and degradation that hinder the algorithm's ability to find the best in the later stage, a dynamic feedback horizon and an adaptive step size are introduced. On the other hand, to ensure that the planned paths are continuous in both orientation and curvature, the Bessel curve theory is introduced to smooth the planned paths. This is demonstrated through a simulation that shows the improved artificial fish swarm algorithm achieving 100% planning accuracy, while ensuring the shortest average path in the same grid environment. Additionally, the smoothed path is continuous in both orientation and curvature, which satisfies the kinematic characteristics of the mobile robot.
Journal Article
Socially situated artificial intelligence enables learning from human interaction
Regardless of how much data artificial intelligence agents have available, agents will inevitably encounter previously unseen situations in real-world deployments. Reacting to novel situations by acquiring new information from other people—socially situated learning—is a core faculty of human development. Unfortunately, socially situated learning remains an open challenge for artificial intelligence agents because they must learn how to interact with people to seek out the information that they lack. In this article, we formalize the task of socially situated artificial intelligence—agents that seek out new information through social interactions with people—as a reinforcement learning problem where the agent learns to identify meaningful and informative questions via rewards observed through social interaction. We manifest our framework as an interactive agent that learns how to ask natural language questions about photos as it broadens its visual intelligence on a large photo-sharing social network. Unlike activelearning methods, which implicitly assume that humans are oracles willing to answer any question, our agent adapts its behavior based on observed norms of which questions people are or are not interested to answer. Through an 8-mo deployment where our agent interacted with 236,000 social media users, our agent improved its performance at recognizing new visual information by 112%. A controlled field experiment confirmed that our agent outperformed an active-learning baseline by 25.6%.This work advances opportunities for continuously improving artificial intelligence (AI) agents that better respect norms in open social environments.
Journal Article
Advances, challenges and opportunities in creating data for trustworthy AI
As artificial intelligence (AI) transitions from research to deployment, creating the appropriate datasets and data pipelines to develop and evaluate AI models is increasingly the biggest challenge. Automated AI model builders that are publicly available can now achieve top performance in many applications. In contrast, the design and sculpting of the data used to develop AI often rely on bespoke manual work, and they critically affect the trustworthiness of the model. This Perspective discusses key considerations for each stage of the data-for-AI pipeline—starting from data design to data sculpting (for example, cleaning, valuation and annotation) and data evaluation—to make AI more reliable. We highlight technical advances that help to make the data-for-AI pipeline more scalable and rigorous. Furthermore, we discuss how recent data regulations and policies can impact AI.
It has become rapidly clear in the past few years that the creation, use and maintenance of high-quality annotated datasets for robust and reliable AI applications requires careful attention. This Perspective discusses challenges, considerations and best practices for various stages in the data-to-AI pipeline, to encourage a more data-centric approach.
Journal Article
Topological light-trapping on a dislocation
Topological insulators have unconventional gapless edge states where disorder-induced back-scattering is suppressed. In photonics, such edge states lead to unidirectional waveguides which are useful for integrated photonic circuitry. Cavity modes, another type of fundamental component in photonic chips, however, are not protected by band topology because of their lower dimensions. Here we demonstrate that concurrent wavevector space and real-space topology, dubbed as dual-topology, can lead to light-trapping in lower dimensions. The resultant photonic-bound state emerges as a Jackiw–Rebbi soliton mode localized on a dislocation in a two-dimensional photonic crystal, as proposed theoretically and discovered experimentally. Such a strongly confined cavity mode is found to be robust against perturbations. Our study unveils a mechanism for topological light-trapping in lower dimensions, which is invaluable for fundamental physics and various applications in photonics.
Although topological confinement of waves to the edges is common, lower-dimensional wave confinement is scarce. Here, Li et al. demonstrate that concurrent wavevector and real-space topology can lead to a topologically protected zero-dimensional cavity mode in a two-dimensional photonic crystal.
Journal Article
Bulk–disclination correspondence in topological crystalline insulators
Most natural and artificial materials have crystalline structures from which abundant topological phases emerge
1
–
6
. However, the bulk–edge correspondence—which has been widely used in experiments to determine the band topology from edge properties—is inadequate in discerning various topological crystalline phases
7
–
16
, leading to challenges in the experimental classification of the large family of topological crystalline materials
4
–
6
. It has been theoretically predicted that disclinations—ubiquitous crystallographic defects—can provide an effective probe of crystalline topology beyond edges
17
–
19
, but this has not yet been confirmed in experiments. Here we report an experimental demonstration of bulk–disclination correspondence, which manifests as fractional spectral charge and robust bound states at the disclinations. The fractional disclination charge originates from the symmetry-protected bulk charge patterns—a fundamental property of many topological crystalline insulators (TCIs). Furthermore, the robust bound states at disclinations emerge as a secondary, but directly observable, property of TCIs. Using reconfigurable photonic crystals as photonic TCIs with higher-order topology, we observe these hallmark features via pump–probe and near-field detection measurements. It is shown that both the fractional charge and the localized states emerge at the disclination in the TCI phase but vanish in the trivial phase. This experimental demonstration of bulk–disclination correspondence reveals a fundamental phenomenon and a paradigm for exploring topological materials.
It is experimentally shown that topological states exist at crystallographic defects in the bulk and that disclination defects trap fractional charges characteristic of topological crystalline insulators.
Journal Article