Dimensional Latent Space Research Articles

Multitask Machine Learning of Collective Variables for Enhanced Sampling of Rare Events.

Computing accurate reaction rates is a central challenge in computational chemistry and biology because of the high cost of free energy estimation with unbiased molecular dynamics. In this work, a data-driven machine learning algorithm is devised to learn collective variables with a multitask neural network, where a common upstream part reduces the high dimensionality of atomic configurations to a low dimensional latent space and separate downstream parts map the latent space to predictions of basin class labels and potential energies. The resulting latent space is shown to be an effective low-dimensional representation, capturing the reaction progress and guiding effective umbrella sampling to obtain accurate free energy landscapes. This approach is successfully applied to model systems including a 5D Müller Brown model, a 5D three-well model, the alanine dipeptide in vacuum, and an Au(110) surface reconstruction unit reaction. It enables automated dimensionality reduction for energy controlled reactions in complex systems, offers a unified and data-efficient framework that can be trained with limited data, and outperforms single-task learning approaches, including autoencoders.

Subject independent emotion recognition using EEG signals employing attention driven neural networks

Electroencephalogram (EEG) based emotional analysis has been employed in medical science, security and human–computer interaction with good success. In the recent past, deep learning-based approaches have significantly improved the classification accuracy when compared to classical signal processing and machine learning based frameworks. But most of them were subject-dependent studies which were not able to generalize on the subject-independent tasks due to the inter-subject variability present in EEG data. In this work, a novel deep learning framework capable of doing subject-independent emotion recognition is presented, consisting of two parts. First, an unsupervised Long Short-Term Memory (LSTM) with channel-attention autoencoder is proposed for getting a subject-invariant latent vector subspace i.e., intrinsic variables present in the EEG data of each individual. Secondly, a convolutional neural network (CNN) with attention framework is presented for performing the task of subject-independent emotion recognition on the encoded lower dimensional latent space representations obtained from the proposed LSTM with channel-attention autoencoder. With the attention mechanism, the proposed approach could highlight the significant time-segments of the EEG signal, which contributes to the emotion under consideration as validated by the results. The proposed approach has been validated using publicly available datasets for EEG signals such as DEAP dataset, SEED dataset and CHB-MIT dataset. With the proposed methodology, average subject independent accuracies of 65.9%, 69.5% for valence and arousal classification in the DEAP dataset and 76.7% for positive–negative classification in SEED dataset are obtained. Further for the CHB-MIT dataset, average subject independent accuracies of 69.1%, 67.6%, 72.3% for Pre-Ictal Vs Ictal, Inter-Ictal Vs Ictal, Pre-Ictal Vs Inter-Ictal classification are obtained, which are state-of-the-art to the best of our knowledge. The proposed end-to-end deep learning framework removes the requirement of different hand engineered features and provides a single comprehensive task agnostic EEG analysis tool capable of performing various kinds of EEG analysis on subject independent data.

A visualization of mechanical Kansei using the DNN approach

It is illustrated in general that the decision-making in conceptual design is highly dependent on the designer's aesthetic sense. However, even in a subjective phase, objective aspects such as mechanical rationality are implicitly or explicitly considered, and these two perspectives are inseparable. A concept of mechanical Kansei defined as “the capability to evaluate and judge impressions in terms of unconscious, intuitive, and integrative information in mechanics,” has been proposed to elucidate how this capability is utilized for design. In this article, we focus on the capability to unconsciously recall the force flow inside a structure; and take the topology-optimized structures as the representation of such force flow. Through learning the structural shapes with a variational autoencoder (VAE), dimensionality compression from the large design variable space to the lower dimensional latent variable space is performed. As a result, it was shown that the two-dimensional latent space reflected the geometric positions of randomly located supports on the design domain. The obtained mappings from the structural conditions to the representation of force flow in structure within the encoder and decoder of VAE can help to understand the mechanical Kansei.

Beets or Cotton? Blind Extraction of Fine Agricultural Classes Using a Convolutional Autoencoder Applied to Temporal SAR Signatures

We present a fully unsupervised learning pipeline, which involves both a projection method and a clustering algorithm dedicated to the pixel-wise classification of multitemporal SAR images. We design a Convolutional Autoencoder as the method to project our time series onto a lower dimensional latent space, where semantically similar temporal signals are placed close together. The additional use of convolutional layers as feature extraction steps allows us to exploit the sequential nature of time series, exhibiting higher representation performance than fully connected layers. The extracted clusters can encapture different semantic levels to either separate classes or extract outlying temporal signals. The application of this method to crop-types mapping enables the extraction of major crop-types within a scene, without supervision. In a labeled context, this method also allows for the extraction of outlying profiles which can lead to the discovery of mislabeled time series.

Probabilistic Autoencoder Using Fisher Information.

Neural networks play a growing role in many scientific disciplines, including physics. Variational autoencoders (VAEs) are neural networks that are able to represent the essential information of a high dimensional data set in a low dimensional latent space, which have a probabilistic interpretation. In particular, the so-called encoder network, the first part of the VAE, which maps its input onto a position in latent space, additionally provides uncertainty information in terms of variance around this position. In this work, an extension to the autoencoder architecture is introduced, the FisherNet. In this architecture, the latent space uncertainty is not generated using an additional information channel in the encoder but derived from the decoder by means of the Fisher information metric. This architecture has advantages from a theoretical point of view as it provides a direct uncertainty quantification derived from the model and also accounts for uncertainty cross-correlations. We can show experimentally that the FisherNet produces more accurate data reconstructions than a comparable VAE and its learning performance also apparently scales better with the number of latent space dimensions.

Tracing individual trajectory of longitudinal tau spreading in deep latent space

AbstractBackgroundAlzheimer’s disease (AD) usually appears in mid‐60s (Late‐Onset AD, LOAD), but also rarely appears in patients younger than 65 (Early‐Onset AD, EOAD), while showing distinct features. Tau shows high correlation with AD prognosis and can be detected through positron emission tomography (PET) scan. Tau data mapped into a low dimensional latent space will help detect longitudinal change pattern along AD spectrum. Furthermore, fitting tensors using individual line segments will enable us to track individual tau trajectory in the latent space.MethodWe recruited 74 healthy controls, 75 mild cognitive impairment patients, and 38 AD patients with two PET scans ([18F]Flortaucipir for tau and [18F]Florbetaben for Aβ) with two‐year follow‐up. We used standardized uptake value ratios (SUVRs) for tau PET data. All patients’ tau data went through autoencoder and principal component analysis, pretrained using baseline (BL) tau data. Using only amyloid positive AD patients’ BL – follow‐up (FU) line segments, we fit tensors and tracked individual trajectories in the latent space using a tensor tracing algorithm.ResultIn the 2‐dimensional latent space the points dispersed toward the upper‐ and right‐side as the disease deteriorated. Reconstructed SUVR maps showed tau accumulation pattern of EOAD through x‐axis, such as tau spreading in temporal, prefrontal, and parietal lobes. Reconstructed map of 1st quartile group in y‐axis showed tau spreading dominance in parietal lobe and 4th quartile group showed dominance in prefrontal, temporal lobe. Each patient’s trajectory, predicted based on the latent space, showed reconstructed map following individual tau accumulation pattern. In addition, EOAD tau spreading trajectory tend to precede LOAD tau spreading trajectory.ConclusionWe decomposed high dimensional tau accumulation pattern into 2 independent components through deep‐learning methods. X‐axis showed EOAD tau accumulation pattern and y‐axis was divided into LOAD tau accumulation pattern and non‐prefrontal dominant pattern. In addition, tensor fitting of the longitudinal vector field enables us to find individual tau spreading trajectories which can be used to predict future and past spreading. Even though this study was done using only one factor, tau, it has the potential to be expanded to other factors such as amyloid or neurodegeneration data.

Adaptive machine learning for time-varying systems: low dimensional latent space tuning

Machine learning (ML) tools such as encoder-decoder convolutional neural networks (CNN) can represent incredibly complex nonlinear functions which map between combinations of images and scalars. For example, CNNs can be used to map combinations of accelerator parameters and images which are 2D projections of the 6D phase space distributions of charged particle beams as they are transported between various particle accelerator locations. Despite their strengths, applying ML to time-varying systems, or systems with shifting distributions, is an open problem, especially for large systems for which collecting new data for re-training is impractical or interrupts operations. Particle accelerators are one example of large time-varying systems for which collecting detailed training data requires lengthy dedicated beam measurements which may no longer be available during regular operations. We present a novel method of adaptive ML for time-varying systems. Our approach is to map very high (N ≈ 100k) dimensional inputs (a combination of scalar parameters and images) into the low dimensional (N ≈ 2) latent space at the output of the encoder section of an encoder-decoder CNN. We then actively tune the low dimensional latent space-based representation of complex system dynamics by the addition of an adaptively tuned feedback vector directly before the decoder sections builds back up to our image-based high-dimensional phase space density representations. This method allows us to learn correlations within and to quickly tune the characteristics of incredibly large parameter space systems and to track their evolution in real time based on feedback without massive new data sets for re-training. We demonstrate that our method can accurately predict and track the phase space of charged particle beams at various locations in a particle accelerator by adaptively adjusting in real-time while the unknown input beam distribution of the accelerator is changing in shape, charge, and offset and while the RF system of the accelerator itself is also changing in an unpredictable way. For FACET-II we demonstrate that such an approach has the potential to use transverse deflecting cavity and energy spread spectrum beam measurements to accurately predict 2D projections of the 6D phase space of the electron beam at the plasma wakefield acceleration interaction point where such diagnostics are unavailable.

Plant-wide troubleshooting and diagnosis using dynamic embedded latent feature analysis

Plant-wide process data are usually high dimensional with dynamics residing in a reduced dimensional latent space. In this paper, we propose a novel procedure for diagnosing and troubleshooting plant-wide process anomalies using dynamic embedded latent feature analysis (DELFA). To remove the impact of external disturbances or exogenous variables, a dynamic inner canonical correlation analysis algorithm with exogenous variables is proposed. Composite loadings and composite weights are derived and applied for diagnosing a feature that is contained in several latent variables. The dynamic embedded latent features are usually related to poor control performance or malfunctioning control instrumentation. The proposed DELFA procedure with dynamic latent scores and composite loadings is applied to two industrial datasets of a chemical plant before and after a troubled control valve was fixed. The case study demonstrates convincingly that latent dynamic features are powerful for troubleshooting of process anomalies and diagnosing their causes in a plant-wide setting.

Dual discriminative auto-encoder network for zero shot image recognition

Zero Shot learning (ZSL) aims to use the information of seen classes to recognize unseen classes, which is achieved by transferring knowledge of the seen classes from the semantic embeddings. Since the domains of the seen and unseen classes do not overlap, most ZSL algorithms often suffer from domain shift problem. In this paper, we propose a Dual Discriminative Auto-encoder Network (DDANet), in which visual features and semantic attributes are self-encoded by using the high dimensional latent space instead of the feature space or the low dimensional semantic space. In the embedded latent space, the features are projected to both preserve their original semantic meanings and have discriminative characteristics, which are realized by applying dual semantic auto-encoder and discriminative feature embedding strategy. Moreover, the cross modal reconstruction is applied to obtain interactive information. Extensive experiments are conducted on four popular datasets and the results demonstrate the superiority of this method.

Deep feature learning of in-cylinder flow fields to analyze cycle-to-cycle variations in an SI engine

Machine learning (ML) models based on a large data set of in-cylinder flow fields of an IC engine obtained by high-speed particle image velocimetry allow the identification of relevant flow structures underlying cycle-to-cycle variations of engine performance. To this end, deep feature learning is employed to train ML models that predict cycles of high and low in-cylinder maximum pressure. Deep convolutional autoencoders are self-supervised-trained to encode flow field features in low dimensional latent space. Without the limitations ascribable to manual feature engineering, ML models based on these learned features are able to classify high energy cycles already from the flow field during late intake and the compression stroke as early as 290 crank angle degrees before top dead center ([Formula: see text]) with a mean accuracy above chance level. The prediction accuracy from [Formula: see text] to [Formula: see text] is comparable to baseline ML approaches utilizing an extensive set of engineered features. Relevant flow structures in the compression stroke are revealed by feature analysis of ML models and are interpreted using conditional averaged flow quantities. This analysis unveils the importance of the horizontal velocity component of in-cylinder flows in predicting engine performance. Combining deep learning and conventional flow analysis techniques promises to be a powerful tool for ultimately revealing high-level flow features relevant to the prediction of cycle-to-cycle variations and further engine optimization.

Towards physical interaction-based sequential mobility assistance using latent generative model of movement state

ABSTRACT In this paper, we address sequential mobility assistance for daily elderly care through physical human–robot interaction. The goal of this work is to develop a robotic assistive system to provide physical support in daily life such as movement transition, e.g. sit-to-stand and walking. Using a mobile human support robotic platform, we propose an unsupervised learning-based approach to providing desirable physical support through an adaptive impedance parameter selection strategy according to the recognized user's movement state in an online manner. Using a latent generative model with a long short-term memory-based variational autoencoder, we first estimate the probability of the user's current movement state based on the sensory information in a low dimensional latent space. Then, the desired impedance parameters are selected adaptively according to the estimated movement state. One of the benefits of such an unsupervised learning approach is that no labeling is necessary in the training phase. Furthermore, our proposed framework is capable of detecting possible novel states such as falling over based on the obtained latent space. In order to demonstrate the proof of concept of our proposed approach, we present the experimental results of performance evaluations of online movement state recognition as well as novel movement detection.

Finding, visualizing, and quantifying latent structure across diverse animal vocal repertoires.

Animals produce vocalizations that range in complexity from a single repeated call to hundreds of unique vocal elements patterned in sequences unfolding over hours. Characterizing complex vocalizations can require considerable effort and a deep intuition about each species' vocal behavior. Even with a great deal of experience, human characterizations of animal communication can be affected by human perceptual biases. We present a set of computational methods for projecting animal vocalizations into low dimensional latent representational spaces that are directly learned from the spectrograms of vocal signals. We apply these methods to diverse datasets from over 20 species, including humans, bats, songbirds, mice, cetaceans, and nonhuman primates. Latent projections uncover complex features of data in visually intuitive and quantifiable ways, enabling high-powered comparative analyses of vocal acoustics. We introduce methods for analyzing vocalizations as both discrete sequences and as continuous latent variables. Each method can be used to disentangle complex spectro-temporal structure and observe long-timescale organization in communication.

Generating Game Levels for Multiple Distinct Games with a Common Latent Space

Generative adversarial networks (GANs) are showing significant promise for procedural content generation (PCG) of game levels. GAN models generate game levels by mapping a low dimensional latent space to game levels in the game space. An intriguing challenge in GAN-based PCG is enabling GANs to produce game levels for multiple distinct games with similar gameplay characteristics using a common underlying low-dimensional representation. In this paper, we present a method for training a novel GAN-based PCG architecture that generates levels in multiple distinct games, starting from a common gameplay action sequence. We evaluate the solvability of the generated games using an automated playing agent and show how the generated game levels are separate representations of the same gameplay by quantifying the similarity between the solution action sequences for the game levels. By probing the common latent space, we show how our approach provides control over the levels generated in distinct games for the presence of desired gameplay patterns in the generated game levels. Results also demonstrate that the GAN-based PCG approach creates novel game levels in multiple distinct games, as indicated by the distance between the action sequences required to solve the game levels.

Generating Game Levels for Multiple Distinct Games with a Common Latent Space

Generative adversarial networks (GANs) are showing significant promise for procedural content generation (PCG) of game levels. GAN models generate game levels by mapping a low dimensional latent space to game levels in the game space. An intriguing challenge in GAN-based PCG is enabling GANs to produce game levels for multiple distinct games with similar gameplay characteristics using a common underlying low-dimensional representation. In this paper, we present a method for training a novel GAN-based PCG architecture that generates levels in multiple distinct games, starting from a common gameplay action sequence. We evaluate the solvability of the generated games using an automated playing agent and show how the generated game levels are separate representations of the same gameplay by quantifying the similarity between the solution action sequences for the game levels. By probing the common latent space, we show how our approach provides control over the levels generated in distinct games for the presence of desired gameplay patterns in the generated game levels. Results also demonstrate that the GAN-based PCG approach creates novel game levels in multiple distinct games, as indicated by the distance between the action sequences required to solve the game levels.

Projection to Latent Spaces Disentangles Pathological Effects on Brain Morphology in the Asymptomatic Phase of Alzheimer's Disease

Alzheimer's disease (AD) continuum is defined as a cascade of several neuropathological processes that can be measured using biomarkers, such as cerebrospinal fluid (CSF) levels of Aβ, p-tau, and t-tau. In parallel, brain anatomy can be characterized through imaging techniques, such as magnetic resonance imaging (MRI). In this work we relate both sets of measurements and seek associations between biomarkers and the brain structure that can be indicative of AD progression. The goal is to uncover underlying multivariate effects of AD pathology on regional brain morphological information. For this purpose, we used the projection to latent structures (PLS) method. Using PLS, we found a low dimensional latent space that best describes the covariance between both sets of measurements on the same subjects. Possible confounder effects (age and sex) on brain morphology are included in the model and regressed out using an orthogonal PLS model. We looked for statistically significant correlations between brain morphology and CSF biomarkers that explain part of the volumetric variance at each region-of-interest (ROI). Furthermore, we used a clustering technique to discover a small set of CSF-related patterns describing the AD continuum. We applied this technique to the study of subjects in the whole AD continuum, from the pre-clinical asymptomatic stages all the way through to the symptomatic groups. Subsequent analyses involved splitting the course of the disease into diagnostic categories: cognitively unimpaired subjects (CU), mild cognitively impaired subjects (MCI), and subjects with dementia (AD-dementia), where all symptoms were due to AD.

Bayesian estimation of the latent dimension and communities in stochastic blockmodels

Spectral embedding of adjacency or Laplacian matrices of undirected graphs is a common technique for representing a network in a lower dimensional latent space, with optimal theoretical guarantees. The embedding can be used to estimate the community structure of the network, with strong consistency results in the stochastic blockmodel framework. One of the main practical limitations of standard algorithms for community detection from spectral embeddings is that the number of communities and the latent dimension of the embedding must be specified in advance. In this article, a novel Bayesian model for simultaneous and automatic selection of the appropriate dimension of the latent space and the number of blocks is proposed. Extensions to directed and bipartite graphs are discussed. The model is tested on simulated and real world network data, showing promising performance for recovering latent community structure.

Database development and exploration of process–microstructure relationships using variational autoencoders

The paper demonstrates graphical representation of a large database containing process–microstructure relationships using an unsupervised machine learning algorithm. Correlating microstructural features to processing is an essential first step to answer the difficult problem of process sequence design. In this paper, a large database of 346,200 orientation distribution functions resulting from a variety of process sequences is constructed, where each sequence comprises up to four stages of tension, compression and rolling along different directions in various permutations. This open-source database is constructed for collaborative development of process design algorithms. The paper demonstrates a novel application of the large database: graphical representation of texture–process relationships. A variational autoencoder is used to reduce the entire database to a two dimensional latent space where variations in processes and properties can be visualized. Using proximity analysis in this latent space, we can quickly unearth multiple process solutions to the problem of texture or property design.

Goal-Directed Planning for Habituated Agents by Active Inference Using a Variational Recurrent Neural Network.

It is crucial to ask how agents can achieve goals by generating action plans using only partial models of the world acquired through habituated sensory-motor experiences. Although many existing robotics studies use a forward model framework, there are generalization issues with high degrees of freedom. The current study shows that the predictive coding (PC) and active inference (AIF) frameworks, which employ a generative model, can develop better generalization by learning a prior distribution in a low dimensional latent state space representing probabilistic structures extracted from well habituated sensory-motor trajectories. In our proposed model, learning is carried out by inferring optimal latent variables as well as synaptic weights for maximizing the evidence lower bound, while goal-directed planning is accomplished by inferring latent variables for maximizing the estimated lower bound. Our proposed model was evaluated with both simple and complex robotic tasks in simulation, which demonstrated sufficient generalization in learning with limited training data by setting an intermediate value for a regularization coefficient. Furthermore, comparative simulation results show that the proposed model outperforms a conventional forward model in goal-directed planning, due to the learned prior confining the search of motor plans within the range of habituated trajectories.

Multi-source information fusion based heterogeneous network embedding

Heterogeneous network embedding aims to learn a mapping between network data in original topological space and vectored data in low dimensional latent space, while encoding valuable information, such as structural and semantic information. The resulting vector representation has shown promising performance for extensive real-world applications, such as node classification and node clustering. However, most of existing methods merely focus on modeling network structural information, ignoring the rich multi-source information of different types of nodes. In this paper, we propose a novel Multi-source Information Fusion based Heterogeneous Network Embedding (MIFHNE) approach. We first capture the semantic information using the strategy of meta-graph based random walk. Subsequently, we jointly model the structural proximity, attribute information and label information in the framework of Nonnegative Matrix Factorization (NMF). Theoretical proofs and comprehensive experiments on two real-world heterogeneous network datasets demonstrate the feasibility and effectiveness of our approach.

Exponential Family Graph Embeddings

Representing networks in a low dimensional latent space is a crucial task with many interesting applications in graph learning problems, such as link prediction and node classification. A widely applied network representation learning paradigm is based on the combination of random walks for sampling context nodes and the traditional Skip-Gram model to capture center-context node relationships. In this paper, we emphasize on exponential family distributions to capture rich interaction patterns between nodes in random walk sequences. We introduce the generic exponential family graph embedding model, that generalizes random walk-based network representation learning techniques to exponential family conditional distributions. We study three particular instances of this model, analyzing their properties and showing their relationship to existing unsupervised learning models. Our experimental evaluation on real-world datasets demonstrates that the proposed techniques outperform well-known baseline methods in two downstream machine learning tasks.