We present two techniques that are shown to yield improved Keyword Spotting (KWS) performance when using the ATWV/MTWV performance measures: (i) score normalization, where the scores of different keywords become commensurate with each other and they more closely correspond to the probability of being correct than raw posteriors; and (ii) system combination, where the detections of multiple systems are merged together, and their scores are interpolated with weights which are optimized using MTWV as the maximization criterion. Both score normalization and system combination approaches show that significant gains in ATWV/MTWV can be obtained, sometimes on the order of 8-10 points (absolute), in five different languages. A variant of these methods resulted in the highest performance for the official surprise language evaluation for the IARPA-funded Babel project in April 2013.
We show how to design secure authentication protocols for a non-standard class of scenarios. In these authentication is not bootstrapped from a PKI, shared secrets or trusted third parties, but rather using a minimum of work by human user(s) implementing the low-bandwidth unspoofable channels between them. We develop both pairwise and group protocols which are essentially optimal in human effort and, given that, computation. We compare our protocols with recent pairwise protocols proposed by, for example, Hoepman and Vaudenay. We introduce and analyse a new cryptographic primitive-a digest function-that is closely related to short-output universal hash functions.
One of the main challenges in pervasive computing is how we can establish secure communication over an untrusted high-bandwidth network without any initial knowledge or a Public Key Infrastructure. An approach studied by a number of researchers is building security though human work creating a low-bandwidth empirical (or authentication) channel where the transmitted information is authentic and cannot be faked or modified. In this paper, we give an analytical survey of authentication protocols of this type. We start with non-interactive authentication schemes, and then move on to analyse a number of strategies used to build interactive pair-wise and group protocols that minimise the human work relative to the amount of security obtained as well as optimising the computation processing. In studying these protocols, we will discover that their security is underlined by the idea of commitment before knowledge, which is refined by two protocol design principles introduced in this survey.
Precisely forecasting wind speed is essential for wind power producers and grid operators. However, this task is challenging due to the stochasticity of wind speed. To accurately predict short-term wind speed under uncertainties, this paper proposed a multi-variable stacked LSTMs model (MSLSTM). The proposed method utilizes multiple historical meteorological variables, such as wind speed, temperature, humidity, pressure, dew point and solar radiation to accurately predict wind speeds. The prediction performance is extensively assessed using real data collected in West Texas, USA. The experimental results show that the proposed MSLSTM can preferably capture and learn uncertainties while output competitive performance.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.