Sum-product networks: A new deep architecture

Poon, Hoifung; Domingos, Pedro

doi:10.1109/iccvw.2011.6130310

Cited by 383 publications

(713 citation statements)

References 23 publications

Supporting

Mentioning

712

Contrasting

Order By: Relevance

“…In the case where an exclusive disjunction is necessary, this requires to replace the classification mechanism (based on W in the proposed network) by a more complex one. Sum-product networks [69] could be an interesting source of inspiration for this replacement.…”

Section: Perspectives and Future Workmentioning

confidence: 99%

Deep unsupervised network for multimodal perception, representation and classification

Droniou

Ivaldi

Sigaud

2015

Robotics and Autonomous Systems

View full text Add to dashboard Cite

In this paper, we tackle the problem of multimodal learning for autonomous robots. Autonomous robots interacting with humans in an evolving environment need the ability to acquire knowledge from their multiple perceptual channels in an unsupervised way. Most of the approaches in the literature exploit engineered methods to process each perceptual modality. In contrast, robots should be able to acquire their own features from the raw sensors, leveraging the information elicited by interaction with their environment: learning from their sensorimotor experience would result in a more efficient strategy in a life-long perspective. To this end, we propose an architecture based on deep networks, which is used by the humanoid robot iCub to learn a task from multiple perceptual modalities (proprioception, vision, audition). By structuring high-dimensional, multimodal information into a set of distinct sub-manifolds in a fully unsupervised way, it performs a substantial dimensionality reduction by providing both a symbolic representation of data and a fine discrimination between two similar stimuli. Moreover, the proposed network is able to exploit multimodal correlations to improve the representation of each modality alone.

show abstract

Section: Perspectives and Future Workmentioning

confidence: 99%

Deep unsupervised network for multimodal perception, representation and classification

Droniou

Ivaldi

Sigaud

2015

Robotics and Autonomous Systems

View full text Add to dashboard Cite

show abstract

“…Using the differential approach introduced in [4], inference is also conceptional easy in these models. In this paper, we consider sum-product networks (SPNs) introduced in [6]. SPNs can be interpreted as Bayesian networks with a deep hierarchical structure of latent variables with a high degree of context-specific independence.…”

Section: Introductionmentioning

confidence: 99%

“…In [4,5,6] and related work, novel types of probabilistic models emerged which allow to control the inference cost during learning but still modeling complex variable dependencies. Using the differential approach introduced in [4], inference is also conceptional easy in these models.…”

Section: Introductionmentioning

confidence: 99%

Modeling speech with sum-product networks: Application to bandwidth extension

Peharz

Kapeller

Mowlaee

et al. 2014

2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

Sum-product networks (SPNs) are a recently proposed type of probabilistic graphical models allowing complex variable interactions while still granting efficient inference. In this paper we demonstrate the suitability of SPNs for modeling log-spectra of speech signals using the application of artificial bandwidth extension, i.e. artificially replacing the high-frequency content which is lost in telephone signals. We use SPNs as observation models in hidden Markov models (HMMs), which model the temporal evolution of log short-time spectra. Missing frequency bins are replaced by the SPNs using most-probable-explanation inference, where the state-dependent reconstructions are weighted with the HMM state posterior. According to subjective listening and objective evaluation, our system consistently and significantly improves the state of the art.

show abstract

“…When it comes to the use of deep learning for generation tasks, we can find various models, such as a deep Boltzmann machines (DBM) [8], [9], a denoising auto-encoder (DAE) [10], a shape Boltzmann machine (ShapeBM) [11], and a sum-product network (SPN) [12]. These models were mainly introduced to capture high-order abstractions for good representation of the observations, rather than for discriminative goal.…”

Section: Introductionmentioning

confidence: 99%

“…In the image generation experiments, we obtained much more realistic images generated from the DRM more than those from the other generative models. [11], and a sum-product network (SPN) [12]. These models were mainly introduced to capture high-order abstractions for good representation of the observations, rather than for discriminative goal.…”

mentioning

confidence: 99%

Deep Relational Model: A Joint Probabilistic Model with a Hierarchical Structure for Bidirectional Estimation of Image and Labels

Nakashika

2018

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

SUMMARYTwo different types of representations, such as an image and its manually-assigned corresponding labels, generally have complex and strong relationships to each other. In this paper, we represent such deep relationships between two different types of visible variables using an energy-based probabilistic model, called a deep relational model (DRM) to improve the prediction accuracies. A DRM stacks several layers from one visible layer on to another visible layer, sandwiching several hidden layers between them. As with restricted Boltzmann machines (RBMs) and deep Boltzmann machines (DBMs), all connections (weights) between two adjacent layers are undirected. During maximum likelihood (ML) -based training, the network attempts to capture the latent complex relationships between two visible variables with its deep architecture. Unlike deep neural networks (DNNs), 1) the DRM is a totally generative model and 2) allows us to generate one visible variables given the other, and 2) the parameters can be optimized in a probabilistic manner. The DRM can be also finetuned using DNNs, like deep belief nets (DBNs) or DBMs pre-training. This paper presents experiments conduced to evaluate the performance of a DRM in image recognition and generation tasks using the MNIST data set. In the image recognition experiments, we observed that the DRM outperformed DNNs even without fine-tuning. In the image generation experiments, we obtained much more realistic images generated from the DRM more than those from the other generative models. [11], and a sum-product network (SPN) [12]. These models were mainly introduced to capture high-order abstractions for good representation of the observations, rather than for discriminative goal. Once obtaining highlevel abstractions, we can, for instance, remove some noise on the observations, or restore missing parts in the observations.Most of the existing deep-learning approaches focus on extracting high-order abstractions from one variable. In this paper, we try to capture such high-order relationships between two different types of variables based on deep learning. For that, we introduced a probabilistic model called a deep relational model (DRM) [13]. A DRM is similar to an RBM and a DBM, each of which is a probabilistic model based on an energy function. The model sandwiches several hidden layers * * between two visible layers and defines a joint probability for the two visible variables. Every two adjacent layers are connected with undirected weights, which are estimated so as to maximize the likelihood of the two visible variables. Interestingly, since the DRM is a totally generative model, it allows us not only to apply it to recognition tasks, but to also generate samples of one variable from the other variable. For example, considering that we have two kinds of variables for a hand-written digit image and a one-hot vector of the labels, we can estimate the label by inferring mean-field posteriors given an image (classification task). On the other hand, by inferring posteriors g...

show abstract

Sum-product networks: A new deep architecture

Cited by 383 publications

References 23 publications

Deep unsupervised network for multimodal perception, representation and classification

Deep unsupervised network for multimodal perception, representation and classification

Modeling speech with sum-product networks: Application to bandwidth extension

Deep Relational Model: A Joint Probabilistic Model with a Hierarchical Structure for Bidirectional Estimation of Image and Labels

Contact Info

Product

Resources

About