SRI International
-
An Autoencoder with Bilingual Sparse Features for Improved Statistical Machine Translation
We propose a natural autoencoder that maps all the discrete and overlapping sparse features for each SCFG rule into a continuous vector, so that the information encoded in sparse feature vectors becomes a dense vector that may enjoy more samples during training and avoid overfitting.
-
Computationally-Efficient Endpointing Features for Natural Spoken Interaction with Personal-Assistant Systems
We elicit personal-assistant speech using a recognizer with a dramatically increased endpoint threshold, and find frequent non-final pauses. Based on the new data, we develop low-cost acoustic features to discriminate non-final from final pauses.
-
Dissociation of Preparatory Attention and Response Monitoring Maturation During Adolescence
The objective was to assess development of EEG markers of these strategies and their role in both preparatory attention (contingent negative variation, CNV) and response monitoring (Error Related Negativity, ERN, and Correct Related Negativity, CRN).
-
Synchrony of Anterior Cingulate Cortex and Insular-Striatal Activation Predicts Ambiguity Aversion in Individuals with Low Impulsivity
To test how dACC functional network connectivity would be modulated by uncertainty and differ between individuals, 24 healthy participants underwent functional MRI in 3 sequential runs: 1 resting-state and 2 decision-making task runs. Individuals with lower nonplanning impulsiveness made greater use of a Pass option and avoided uncertain ambiguous situations.
-
Towards Quantifying the Completeness of BDI Goals
We sketch a pragmatic but principled mechanism for quantifying the level of completeness of goals in a Belief-Desire-Intention–like agent.
-
Blended Learning Report
With funding from the Michael & Susan Dell Foundation, SRI’s Center for Technology in Learning studied the adoption of blended learning models in selected schools in California and Louisiana associated with five different charter management organizations during the 2011-12 school year. This research report presents the findings of this formative and summative research effort, including…
-
On the Feasibility of Deploying Cell Anomaly Detection in Operational Cellular Networks
In this paper, we build on our previous work and study the feasibility of an operational deployment of an adaptive ensemble-method framework for modeling cell behavior.
-
Effects of Thermal Treatment on Radiative Properties of HVPE Grown InP Layers
We have studied radiative properties of 21 micron thick InP layers grown by HVPE and found them comparable to those of best luminescent bulk InP virgin wafers. This opens up the possibility of implementing free-standing epitaxial InP scintillator structures endowed with surface photodiodes for registration of the scintillation.
-
Forensic Prescreening System Using Coded Aperture Snapshot Spectral Imager
We present a camera system for instantaneous, non-destructive capture of spectral signatures for forensic analysis. Our system detects highly probative samples in the forensic scene mixed by the multiple target objects by combining a coded aperture snapshot spectral imager with a multi-spectral detection algorithm.
-
Late Fusion and Calibration for Multimedia Event Detection Using Few Examples
In this paper, we present two parametric approaches to late fusion: a normalization scheme for arithmetic mean fusion (logistic averaging) and a fusion scheme based on logistic regression, and compare them to widely used rule-based fusion schemes.
-
Unscented Transform for iVector-Based Noisy Speaker Recognition
In this paper, it is proposed to substitute the first order VTS by an unscented transform, where unlike VTS, the nonlinear function is not applied over the clean model parameters directly, but over a set of sampled points.
-
Effective Use of DCTS for Contextualizing Features for Speaker Recognition
This article proposes a new approach for contextualizing features for speaker recognition through the discrete cosine transform (DCT).