The convergence speed of stochastic gradient descent (SGD) can be improved by actively selecting mini-batches. We explore sampling schemes where similar data points are less likely to be selected in the same mini-batch. In particular, we prove that such repulsive sampling schemes lower the variance of the gradient estimator. This generalizes recent work on using Determinantal Point Processes (DPPs) for mini-batch diversification (Zhang et al., 2017) to the broader class of repulsive point processes. We first show that the phenomenon of variance reduction by diversified sampling generalizes in particular to non-stationary point processes. We then show that other point processes may be computationally much more efficient than DPPs. In particular, we propose and investigate Poisson Disk sampling-frequently encountered in the computer graphics community-for this task. We show empirically that our approach improves over standard SGD both in terms of convergence speed as well as final model performance.

We propose a developmental approach that allows a robot to interpret and describe the actions of human agents by reusing previous experience. The robot first learns the association between words and object affordances by manipulating the objects in its environment. It then uses this information to learn a mapping between its own actions and those performed by a human in a shared environment. It finally fuses the information from these two models to interpret and describe human actions in light of its own experience. In our experiments, we show that the model can be used flexibly to do inference on different aspects of the scene. We can predict the effects of an action on the basis of object properties. We can revise the belief that a certain action occurred, given the observed effects of the human action. In an early action recognition fashion, we can anticipate the effects when the action has only been partially observed. By estimating the probability of words given the evidence and feeding them into a pre-defined grammar, we can generate relevant descriptions of the scene. We believe that this is a step towards providing robots with the fundamental skills to engage in social collaboration with humans.

This study proposes, develops, and evaluates methods for modeling the eye-gaze direction and head orientation of a person in multiparty open-world dialogues, as a function of low-level communicative signals generated by his/hers interlocutors. These signals include speech activity, eye-gaze direction, and head orientation, all of which can be estimated in real time during the interaction. By utilizing these signals and novel data representations suitable for the task and context, the developed methods can generate plausible candidate gaze targets in real time. The methods are based on Feedforward Neural Networks and Long Short-Term Memory Networks. The proposed methods are developed using several hours of unrestricted interaction data and their performance is compared with a heuristic baseline method. The study offers an extensive evaluation of the proposed methods that investigates the contribution of different predictors to the accurate generation of candidate gaze targets. The results show that the methods can accurately generate candidate gaze targets when the person being modeled is in a listening state. However, when the person being modeled is in a speaking state, the proposed methods yield significantly lower performance.

This paper presents a self-supervised method for visual detection of the active speaker in a multi-person spoken interaction scenario. Active speaker detection is a fundamental prerequisite for any artificial cognitive system attempting to acquire language in social settings. The proposed method is intended to complement the acoustic detection of the active speaker, thus improving the system robustness in noisy conditions. The method can detect an arbitrary number of possibly overlapping active speakers based exclusively on visual information about their face. Furthermore, the method does not rely on external annotations, thus complying with cognitive development. Instead, the method uses information from the auditory modality to support learning in the visual domain. This paper reports an extensive evaluation of the proposed method using a large multi-person face-to-face interaction dataset. The results show good performance in a speaker dependent setting. However, in a speaker independent setting the proposed method yields a significantly lower performance. We believe that the proposed method represents an essential component of any artificial cognitive system or robotic platform engaging in social interactions.

This paper presents a self-supervised method for visual detection of the active speaker in a multi-person spoken interaction scenario. Active speaker detection is a fundamental prerequisite for any artificial cognitive system attempting to acquire language in social settings. The proposed method is intended to complement the acoustic detection of the active speaker, thus improving the system robustness in noisy conditions. The method can detect an arbitrary number of possibly overlapping active speakers based exclusively on visual information about their face. Furthermore, the method does not rely on external annotations, thus complying with cognitive development. Instead, the method uses information from the auditory modality to support learning in the visual domain. This paper reports an extensive evaluation of the proposed method using a large multi-person face-to-face interaction dataset. The results show good performance in a speaker dependent setting. However, in a speaker independent setting the proposed method yields a significantly lower performance. We believe that the proposed method represents an essential component of any artificial cognitive system or robotic platform engaging in social interactions.

We analyse the properties of shallow and deep representa-tions of speech. Mel frequency cepstral coefficients (MFCC) are compared to representations learned by a four layer Deep Belief Network (DBN) in terms of discriminative power and invariance to irrelevant factors such as speaker identity or gender. To avoid the influence of supervised statistical modelling, an unsupervised isolated word classification task is used for the comparison. The deep representations are also obtained with unsupervised training (no back-propagation pass is performed). The results show that DBN features provide a more concise clustering and higher match between clusters and word categories in terms of adjusted Rand score. Some of the confusions present with the MFCC features are, however, retained even with the DBN features.

This investigation explores perceptual and acoustic characteristics of children's successful and unsuccessful productions of /t/ and /k/, with a specific aim of exploring perceptual sensitivity to phonetic detail, and the extent to which this sensitivity is reflected in the acoustic domain. Recordings were collected from 4- to 8-year-old children with a speech sound disorder (SSD) who misarticulated one of the target plosives, and compared to productions recorded from peers with typical speech development (TD). Perceptual responses were registered with regards to a visual-analog scale, ranging from "clear [t]" to "clear [k]." Statistical models of prototypical productions were built, based on spectral moments and discrete cosine transform features, and used in the scoring of SSD productions. In the perceptual evaluation, " clear substitutions" were rated as less prototypical than correct productions. Moreover, target-appropriate productions of /t/ and /k/ produced by children with SSD were rated as less prototypical than those produced by TD peers. The acoustical modeling could to a large extent discriminate between the gross categories /t/ and /k/, and scored the SSD utterances on a continuous scale that was largely consistent with the category of production. However, none of the methods exhibited the same sensitivity to phonetic detail as the human listeners.

Repetitions in Spoken Dialogue Systems can be a symptom of problematic communication. Such repetitions are often due to speech recognition errors, which in turn makes it harder to use the output of the speech recognizer to detect repetitions. In this paper, we combine the alignment score obtained using phonetic distances with dialogue-related features to improve repetition detection. To evaluate the method proposed we compare several alignment techniques from edit distance to DTW-based distance, previously used in Spoken-Term detection tasks. We also compare two different methods to compute the phonetic distance: the first one using the phoneme sequence, and the second one using the distance between the phone posterior vectors. Two different datasets were used in this evaluation: a bus-schedule information system (in English) and a call routing system (in Swedish). The results show that approaches using phoneme distances over-perform approaches using Levenshtein distances between ASR outputs for repetition detection.

We present a data set of human activities that includes both visual data (RGB-D video and six Degrees Of Freedom (DOF) object pose estimation) and acoustic data. Our vision is that robots need to merge information from multiple perceptional modalities to operate robustly and autonomously in an unstructured environment.

Humans are able to merge information from multiple perceptional modalities and formulate a coherent representation of the world. Our thesis is that robots need to do the same in order to operate robustly and autonomously in an unstructured environment. It has also been shown in several fields that multiple sources of information can complement each other, overcoming the limitations of a single perceptual modality. Hence, in this paper we introduce a data set of actions that includes both visual data (RGB-D video and 6DOF object pose estimation) and acoustic data. We also propose a method for recognizing and segmenting actions from continuous audio-visual data. The proposed method is employed for extensive evaluation of the descriptive power of the two modalities, and we discuss how they can be used jointly to infer a coherent interpretation of the recorded action.