Author
Ron Weiss
Other affiliations: French Institute for Research in Computer Science and Automation, Google, University of California, San Diego ...read more
Bio: Ron Weiss is an academic researcher from Massachusetts Institute of Technology. The author has contributed to research in topics: Synthetic biology & Speech synthesis. The author has an hindex of 82, co-authored 292 publications receiving 89189 citations. Previous affiliations of Ron Weiss include French Institute for Research in Computer Science and Automation & Google.
Papers published on a yearly basis
Papers
More filters
•
TL;DR: Tacotron 2 as mentioned in this paper uses a recurrent sequence-to-sequence feature prediction network that maps character embeddings to mel-scale spectrograms, followed by a modified WaveNet model acting as a vocoder to synthesize timedomain waveforms.
Abstract: This paper describes Tacotron 2, a neural network architecture for speech synthesis directly from text. The system is composed of a recurrent sequence-to-sequence feature prediction network that maps character embeddings to mel-scale spectrograms, followed by a modified WaveNet model acting as a vocoder to synthesize timedomain waveforms from those spectrograms. Our model achieves a mean opinion score (MOS) of $4.53$ comparable to a MOS of $4.58$ for professionally recorded speech. To validate our design choices, we present ablation studies of key components of our system and evaluate the impact of using mel spectrograms as the input to WaveNet instead of linguistic, duration, and $F_0$ features. We further demonstrate that using a compact acoustic intermediate representation enables significant simplification of the WaveNet architecture.
733 citations
••
TL;DR: In this article, the authors demonstrate that by coupling gene expression to cell survival and death using cell-cell communication, they can program the dynamics of a population despite variability in the behaviour of individual cells.
Abstract: De novo engineering of gene circuits inside cells is extremely difficult, and efforts to realize predictable and robust performance must deal with noise in gene expression and variation in phenotypes between cells. Here we demonstrate that by coupling gene expression to cell survival and death using cell–cell communication, we can programme the dynamics of a population despite variability in the behaviour of individual cells. Specifically, we have built and characterized a ‘population control’ circuit that autonomously regulates the density of an Escherichia coli population. The cell density is broadcasted and detected by elements from a bacterial quorum-sensing system, which in turn regulate the death rate. As predicted by a simple mathematical model, the circuit can set a stable steady state in terms of cell density and gene expression that is easily tunable by varying the stability of the cell–cell communication signal. This circuit incorporates a mechanism for programmed death in response to changes in the environment, and allows us to probe the design principles of its more complex natural counterparts.
704 citations
••
TL;DR: A scalable transcriptional/posttranscriptional synthetic regulatory circuit—a cell-type “classifier”— is shown that selectively identifies HeLa cancer cells and triggers apoptosis without affecting non-HeLa cell types.
Abstract: Engineered biological systems that integrate multi-input sensing, sophisticated information processing, and precisely regulated actuation in living cells could be useful in a variety of applications. For example, anticancer therapies could be engineered to detect and respond to complex cellular conditions in individual cells with high specificity. Here, we show a scalable transcriptional/posttranscriptional synthetic regulatory circuit--a cell-type "classifier"--that senses expression levels of a customizable set of endogenous microRNAs and triggers a cellular response only if the expression levels match a predetermined profile of interest. We demonstrate that a HeLa cancer cell classifier selectively identifies HeLa cells and triggers apoptosis without affecting non-HeLa cell types. This approach also provides a general platform for programmed responses to other complex cell states.
637 citations
••
[...]
TL;DR: Newton’s language Regiment, also a functional language, is designed to gather streams of data from regions of the amorphous computer and accumulate them at a single point, which allows Regiment to provide region-wide summary functions that are difficult to implement in Proto.
Abstract: ion to Continuous Space and Time The amorphous model postulates computing particles distributed throughout a space. If the particles are dense, one can imagine the particles as actually filling the space, and create programming abstractions that view the space itself as the object being programmed, rather than the collection of particles. Beal and Bachrach [10, 4] pursued this approach by creating a language, Proto, where programmers specify the behavior of an amorphous computer as though it were a continuous material filling the space it occupies. Proto programs manipulate fields of values spanning the entire space. Programming primitives are designed to make it simple to compile global operations to operations at each point of the continuum. These operations are approximated by having each device represent a nearby chunk of space. Programs are specified in space and time units that are independent of the distribution of particles and of the particulars of communication and execution on those particles (Figure 5). Programs are composed functionally, and many of the details of communication and composition are made implicit by Proto’s runtime system, allowing complex programs to be expressed simply. Proto has been applied to applications in sensor networks like target tracking and threat avoidance, to swarm robotics and to modular robotics, e.g., generating a planar wave for coordinated actuation. Newton’s language Regiment [45, 44] also takes a continuous view of space and time. Regiment is organized in terms of stream operations, where each stream represents a time-varying quantity over a part of space, for example, the average value of the temperature over a disc of a given radius centered at a designated point. Regiment, also a functional language, is designed to gather streams of data from regions of the amorphous computer and accumulate them at a single point. This assumption allows Regiment to provide region-wide summary functions that are difficult to implement in Proto.
554 citations
••
TL;DR: The construction of synthetic transcriptional cascades comprising one, two, and three repression stages are reported, which enable us to analyze sensitivity and noise propagation as a function of network complexity.
Abstract: The precise nature of information flow through a biological network, which is governed by factors such as response sensitivities and noise propagation, greatly affects the operation of biological systems. Quantitative analysis of these properties is often difficult in naturally occurring systems but can be greatly facilitated by studying simple synthetic networks. Here, we report the construction of synthetic transcriptional cascades comprising one, two, and three repression stages. These model systems enable us to analyze sensitivity and noise propagation as a function of network complexity. We demonstrate experimentally steady-state switching behavior that becomes sharper with longer cascades. The regulatory mechanisms that confer this ultrasensitive response both attenuate and amplify phenotypical variations depending on the system's input conditions. Although noise attenuation allows the cascade to act as a low-pass filter by rejecting short-lived perturbations in input conditions, noise amplification results in loss of synchrony among a cell population. The experimental results demonstrating the above network properties correlate well with simulations of a simple mathematical model of the system.
543 citations
Cited by
More filters
28 Jul 2005
TL;DR: PfPMP1)与感染红细胞、树突状组胞以及胎盘的单个或多个受体作用,在黏附及免疫逃避中起关键的作�ly.
Abstract: 抗原变异可使得多种致病微生物易于逃避宿主免疫应答。表达在感染红细胞表面的恶性疟原虫红细胞表面蛋白1(PfPMP1)与感染红细胞、内皮细胞、树突状细胞以及胎盘的单个或多个受体作用,在黏附及免疫逃避中起关键的作用。每个单倍体基因组var基因家族编码约60种成员,通过启动转录不同的var基因变异体为抗原变异提供了分子基础。
18,940 citations
••
13 Aug 2016TL;DR: XGBoost as discussed by the authors proposes a sparsity-aware algorithm for sparse data and weighted quantile sketch for approximate tree learning to achieve state-of-the-art results on many machine learning challenges.
Abstract: Tree boosting is a highly effective and widely used machine learning method. In this paper, we describe a scalable end-to-end tree boosting system called XGBoost, which is used widely by data scientists to achieve state-of-the-art results on many machine learning challenges. We propose a novel sparsity-aware algorithm for sparse data and weighted quantile sketch for approximate tree learning. More importantly, we provide insights on cache access patterns, data compression and sharding to build a scalable tree boosting system. By combining these insights, XGBoost scales beyond billions of examples using far fewer resources than existing systems.
14,872 citations
••
01 Apr 1998TL;DR: This paper provides an in-depth description of Google, a prototype of a large-scale search engine which makes heavy use of the structure present in hypertext and looks at the problem of how to effectively deal with uncontrolled hypertext collections where anyone can publish anything they want.
Abstract: In this paper, we present Google, a prototype of a large-scale search engine which makes heavy use of the structure present in hypertext. Google is designed to crawl and index the Web efficiently and produce much more satisfying search results than existing systems. The prototype with a full text and hyperlink database of at least 24 million pages is available at http://google.stanford.edu/. To engineer a search engine is a challenging task. Search engines index tens to hundreds of millions of web pages involving a comparable number of distinct terms. They answer tens of millions of queries every day. Despite the importance of large-scale search engines on the web, very little academic research has been done on them. Furthermore, due to rapid advance in technology and web proliferation, creating a web search engine today is very different from three years ago. This paper provides an in-depth description of our large-scale web search engine -- the first such detailed public description we know of to date. Apart from the problems of scaling traditional search techniques to data of this magnitude, there are new technical challenges involved with using the additional information present in hypertext to produce better search results. This paper addresses this question of how to build a practical large-scale system which can exploit the additional information present in hypertext. Also we look at the problem of how to effectively deal with uncontrolled hypertext collections where anyone can publish anything they want.
14,696 citations
•
11 Nov 1999
TL;DR: This paper describes PageRank, a mathod for rating Web pages objectively and mechanically, effectively measuring the human interest and attention devoted to them, and shows how to efficiently compute PageRank for large numbers of pages.
Abstract: The importance of a Web page is an inherently subjective matter, which depends on the readers interests, knowledge and attitudes. But there is still much that can be said objectively about the relative importance of Web pages. This paper describes PageRank, a mathod for rating Web pages objectively and mechanically, effectively measuring the human interest and attention devoted to them. We compare PageRank to an idealized random Web surfer. We show how to efficiently compute PageRank for large numbers of pages. And, we show how to apply PageRank to search and to user navigation.
14,400 citations
••
TL;DR: This paper proposes a novel sparsity-aware algorithm for sparse data and weighted quantile sketch for approximate tree learning and provides insights on cache access patterns, data compression and sharding to build a scalable tree boosting system called XGBoost.
Abstract: Tree boosting is a highly effective and widely used machine learning method. In this paper, we describe a scalable end-to-end tree boosting system called XGBoost, which is used widely by data scientists to achieve state-of-the-art results on many machine learning challenges. We propose a novel sparsity-aware algorithm for sparse data and weighted quantile sketch for approximate tree learning. More importantly, we provide insights on cache access patterns, data compression and sharding to build a scalable tree boosting system. By combining these insights, XGBoost scales beyond billions of examples using far fewer resources than existing systems.
13,333 citations