scholarly journals Flexible working memory through selective gating and attentional tagging

2019 ◽  
Author(s):  
Wouter Kruijne ◽  
Sander M. Bohte ◽  
Pieter R. Roelfsema ◽  
Christian N. L. Olivers
Keyword(s):  
Neural Network ◽  
Working Memory ◽  
Network Architecture ◽  
Sensory Information ◽  
Task Demands ◽  
Control Mechanisms ◽  
Neural Network Models ◽  
Complex Tasks ◽  
Memory Content ◽  
The Brain

AbstractWorking memory is essential for intelligent behavior as it serves to guide behavior of humans and nonhuman primates when task-relevant stimuli are no longer present to the senses. Moreover, complex tasks often require that multiple working memory representations can be flexibly and independently maintained, prioritized, and updated according to changing task demands. Thus far, neural network models of working memory have been unable to offer an integrative account of how such control mechanisms are implemented in the brain and how they can be acquired in a biologically plausible manner. Here, we present WorkMATe, a neural network architecture that models cognitive control over working memory content and learns the appropriate control operations needed to solve complex working memory tasks. Key components of the model include a gated memory circuit that is controlled by internal actions, encoding sensory information through untrained connections, and a neural circuit that matches sensory inputs to memory content. The network is trained by means of a biologically plausible reinforcement learning rule that relies on attentional feedback and reward prediction errors to guide synaptic updates. We demonstrate that the model successfully acquires policies to solve classical working memory tasks, such as delayed match-to-sample and delayed pro-saccade/antisaccade tasks. In addition, the model solves much more complex tasks including the hierarchical 12-AX task or the ABAB ordered recognition task, which both demand an agent to independently store and updated multiple items separately in memory. Furthermore, the control strategies that the model acquires for these tasks subsequently generalize to new task contexts with novel stimuli. As such, WorkMATe provides a new solution for the neural implementation of flexible memory control.Author SummaryWorking Memory, the ability to briefly store sensory information and use it to guide behavior, is a cornerstone of intelligent behavior. Existing neural network models of Working Memory typically focus on how information is stored and maintained in the brain, but do not address how memory content is controlled: how the brain can selectively store only stimuli that are relevant for a task, or how different stimuli can be maintained in parallel, and subsequently replaced or updated independently according to task demands. The models that do implement control mechanisms are typically not trained in a biologically plausible manner, and do not explain how the brain learns such control. Here, we present WorkMATe, a neural network architecture that implements flexible cognitive control and learns to apply these control mechanisms using a biologically plausible reinforcement learning method. We demonstrate that the model acquires control policies to solve a range of both simple and more complex tasks. Moreover, the acquired control policies generalize to new situations, as with human cognition. This way, WorkMATe provides new insights into the neural organization of Working Memory beyond mere storage and retrieval.

scholarly journals Flexible Working Memory Through Selective Gating and Attentional Tagging

2021 ◽  
Vol 33 (1) ◽  
pp. 1-40 ◽  
Author(s):  
Wouter Kruijne ◽  
Sander M. Bohte ◽  
Pieter R. Roelfsema ◽  
Christian N. L. Olivers
Keyword(s):  
Neural Network ◽  
Working Memory ◽  
Network Architecture ◽  
Neural Circuit ◽  
Recognition Task ◽  
Learning Rule ◽  
Task Demands ◽  
Neural Network Architecture ◽  
Complex Tasks ◽  
Memory Content

Working memory is essential: it serves to guide intelligent behavior of humans and nonhuman primates when task-relevant stimuli are no longer present to the senses. Moreover, complex tasks often require that multiple working memory representations can be flexibly and independently maintained, prioritized, and updated according to changing task demands. Thus far, neural network models of working memory have been unable to offer an integrative account of how such control mechanisms can be acquired in a biologically plausible manner. Here, we present WorkMATe, a neural network architecture that models cognitive control over working memory content and learns the appropriate control operations needed to solve complex working memory tasks. Key components of the model include a gated memory circuit that is controlled by internal actions, encoding sensory information through untrained connections, and a neural circuit that matches sensory inputs to memory content. The network is trained by means of a biologically plausible reinforcement learning rule that relies on attentional feedback and reward prediction errors to guide synaptic updates. We demonstrate that the model successfully acquires policies to solve classical working memory tasks, such as delayed recognition and delayed pro-saccade/anti-saccade tasks. In addition, the model solves much more complex tasks, including the hierarchical 12-AX task or the ABAB ordered recognition task, both of which demand an agent to independently store and updated multiple items separately in memory. Furthermore, the control strategies that the model acquires for these tasks subsequently generalize to new task contexts with novel stimuli, thus bringing symbolic production rule qualities to a neural network architecture. As such, WorkMATe provides a new solution for the neural implementation of flexible memory control.

scholarly journals Engineering recurrent neural networks from task-relevant manifolds and dynamics

2019 ◽  
Author(s):  
Eli Pollock ◽  
Mehrdad Jazayeri
Keyword(s):  
Neural Network ◽  
Working Memory ◽  
Recurrent Neural Network ◽  
Network Dynamics ◽  
Memory Task ◽  
Network Models ◽  
Network Activity ◽  
Mathematical Framework ◽  
Neural Network Models ◽  
The Brain

AbstractMany cognitive processes involve transformations of distributed representations in neural populations, creating a need for population-level models. Recurrent neural network models fulfill this need, but there are many open questions about how their connectivity gives rise to dynamics that solve a task. Here, we present a method for finding the connectivity of networks for which the dynamics are specified to solve a task in an interpretable way. We apply our method to a working memory task by synthesizing a network that implements a drift-diffusion process over a ring-shaped manifold. We also use our method to demonstrate how inputs can be used to control network dynamics for cognitive flexibility and explore the relationship between representation geometry and network capacity. Our work fits within the broader context of understanding neural computations as dynamics over relatively low-dimensional manifolds formed by correlated patterns of neurons.Author SummaryNeurons in the brain form intricate networks that can produce a vast array of activity patterns. To support goal-directed behavior, the brain must adjust the connections between neurons so that network dynamics can perform desirable computations on behaviorally relevant variables. A fundamental goal in computational neuroscience is to provide an understanding of how network connectivity aligns the dynamics in the brain to the dynamics needed to track those variables. Here, we develop a mathematical framework for creating recurrent neural network models that can address this problem. Specifically, we derive a set of linear equations that constrain the connectivity to afford a direct mapping of task-relevant dynamics onto network activity. We demonstrate the utility of this technique by creating and analyzing a set of network models that can perform a simple working memory task. We then extend the approach to show how additional constraints can furnish networks whose dynamics are controlled flexibly by external inputs. Finally, we exploit the flexibility of this technique to explore the robustness and capacity limitations of recurrent networks. This network synthesis method provides a powerful means for generating and validating hypotheses about how task-relevant computations can emerge from network dynamics.

Measuring the World

2018 ◽  
Author(s):  
Ann-Sophie Barwich
Keyword(s):  
Sensory Information ◽  
Network Models ◽  
Model System ◽  
Stimulus Input ◽  
Expectancy Effects ◽  
Neural Network Models ◽  
Perceptual Object ◽  
External Objects ◽  
The Common ◽  
The Brain

How much does stimulus input shape perception? The common-sense view is that our perceptions are representations of objects and their features and that the stimulus structures the perceptual object. The problem for this view concerns perceptual biases as responsible for distortions and the subjectivity of perceptual experience. These biases are increasingly studied as constitutive factors of brain processes in recent neuroscience. In neural network models the brain is said to cope with the plethora of sensory information by predicting stimulus regularities on the basis of previous experiences. Drawing on this development, this chapter analyses perceptions as processes. Looking at olfaction as a model system, it argues for the need to abandon a stimulus-centred perspective, where smells are thought of as stable percepts, computationally linked to external objects such as odorous molecules. Perception here is presented as a measure of changing signal ratios in an environment informed by expectancy effects from top-down processes.

Spatial Variability Aware Deep Neural Networks (SVANN): A General Approach

2021 ◽  
Vol 12 (6) ◽  
pp. 1-21
Author(s):  
Jayant Gupta ◽  
Carl Molnar ◽  
Yiqun Xie ◽  
Joe Knight ◽  
Shashi Shekhar
Keyword(s):  
Neural Network ◽  
Spatial Variability ◽  
Network Architecture ◽  
Network Models ◽  
Neural Network Architecture ◽  
Neural Network Models ◽  
Climatic Zones ◽  
The Neural Network ◽  
Plant Hardiness ◽  
Interpretation Model

Spatial variability is a prominent feature of various geographic phenomena such as climatic zones, USDA plant hardiness zones, and terrestrial habitat types (e.g., forest, grasslands, wetlands, and deserts). However, current deep learning methods follow a spatial-one-size-fits-all (OSFA) approach to train single deep neural network models that do not account for spatial variability. Quantification of spatial variability can be challenging due to the influence of many geophysical factors. In preliminary work, we proposed a spatial variability aware neural network (SVANN-I, formerly called SVANN ) approach where weights are a function of location but the neural network architecture is location independent. In this work, we explore a more flexible SVANN-E approach where neural network architecture varies across geographic locations. In addition, we provide a taxonomy of SVANN types and a physics inspired interpretation model. Experiments with aerial imagery based wetland mapping show that SVANN-I outperforms OSFA and SVANN-E performs the best of all.

scholarly journals A data-driven neural network architecture for sentiment analysis

2019 ◽  
Vol 53 (1) ◽  
pp. 2-19 ◽  
Author(s):  
Erion Çano ◽  
Maurizio Morisio
Keyword(s):  
Neural Network ◽  
Sentiment Analysis ◽  
Network Architecture ◽  
Network Models ◽  
Data Sets ◽  
Feature Maps ◽  
Neural Network Architecture ◽  
Neural Network Models ◽  
Content Type ◽  
Max Pooling

Purpose The fabulous results of convolution neural networks in image-related tasks attracted attention of text mining, sentiment analysis and other text analysis researchers. It is, however, difficult to find enough data for feeding such networks, optimize their parameters, and make the right design choices when constructing network architectures. The purpose of this paper is to present the creation steps of two big data sets of song emotions. The authors also explore usage of convolution and max-pooling neural layers on song lyrics, product and movie review text data sets. Three variants of a simple and flexible neural network architecture are also compared. Design/methodology/approach The intention was to spot any important patterns that can serve as guidelines for parameter optimization of similar models. The authors also wanted to identify architecture design choices which lead to high performing sentiment analysis models. To this end, the authors conducted a series of experiments with neural architectures of various configurations. Findings The results indicate that parallel convolutions of filter lengths up to 3 are usually enough for capturing relevant text features. Also, max-pooling region size should be adapted to the length of text documents for producing the best feature maps. Originality/value Top results the authors got are obtained with feature maps of lengths 6–18. An improvement on future neural network models for sentiment analysis could be generating sentiment polarity prediction of documents using aggregation of predictions on smaller excerpt of the entire text.

scholarly journals Data-to-Text Generation with Content Selection and Planning

2019 ◽  
Vol 33 ◽  
pp. 6908-6915 ◽  
Author(s):  
Ratish Puduppully ◽  
Li Dong ◽  
Mirella Lapata
Keyword(s):  
Neural Network ◽  
Network Architecture ◽  
Large Scale ◽  
Network Models ◽  
Text Generation ◽  
Neural Network Models ◽  
Generation Task ◽  
Content Selection ◽  
End To End ◽  
Two Stages

Recent advances in data-to-text generation have led to the use of large-scale datasets and neural network models which are trained end-to-end, without explicitly modeling what to say and in what order. In this work, we present a neural network architecture which incorporates content selection and planning without sacrificing end-to-end training. We decompose the generation task into two stages. Given a corpus of data records (paired with descriptive documents), we first generate a content plan highlighting which information should be mentioned and in which order and then generate the document while taking the content plan into account. Automatic and human-based evaluation experiments show that our model1 outperforms strong baselines improving the state-of-the-art on the recently released RotoWIRE dataset.

scholarly journals A Multi-Task Pipeline with Specialized Streams for Classification and Segmentation of Infection Manifestations in COVID-19 Scans

2020 ◽  
Author(s):  
Ahmad Al-Kabbany ◽  
Shimaa El-bana ◽  
Maha Sharkas
Keyword(s):  
Neural Network ◽  
Network Architecture ◽  
Performance Metrics ◽  
Recent Literature ◽  
Ct Scans ◽  
Computational Time ◽  
Neural Network Models ◽  
Research Attention ◽  
X Ray ◽  
Deep Model

We are concerned with the challenge of coronavirus disease (COVID-19) detection in chest<br>X-ray and Computed Tomography (CT) scans, and the classification and segmentation of related<br>infection manifestations. Even though it is arguably not an established diagnostic tool, using machine<br>learning-based analysis of COVID-19 medical scans has shown the potential to provide a preliminary<br>digital second opinion. This can help in managing the current pandemic, and thus has been attracting<br>significant research attention. In this research, we propose a multi-task pipeline that takes advantage<br>of the growing advances in deep neural network models. In the first stage, we fine-tuned an<br>Inception-v3 deep model for COVID-19 recognition using multi-modal learning, i.e., using X-ray and<br>CT scans. In addition to outperforming other deep models on the same task in the recent literature,<br>with an attained accuracy of 99.4%, we also present comparative analysis for multi-modal learning<br>against learning from X-ray scans alone. The second and the third stages of the proposed pipeline<br>complement one another in dealing with different types of infection manifestations. The former<br>features a convolutional neural network architecture for recognizing three types of manifestations,<br>while the latter transfers learning from another knowledge domain, namely, pulmonary nodule<br>segmentation in CT scans, to produce binary masks for segmenting the regions corresponding to<br>these manifestations. Our proposed pipeline also features specialized streams in which multiple deep<br>models are trained separately to segment specific types of infection manifestations, and we show the<br>significant impact that this framework has on various performance metrics. We evaluate the<br>proposed models on widely adopted datasets, and we demonstrate an increase of approximately 4%<br>and 7% for dice coefficient and mean intersection-over-union (mIoU), respectively, while achieving<br>60% reduction in computational time, compared to the recent literature.<br> <br>

scholarly journals Modeling the online social transmission through the theory of complex adaptive systems and neural network models based on the brain dynamics

2015 ◽  
Author(s):  
Ηλίας Λυμπερόπουλος
Keyword(s):  
Neural Network ◽  
Complex Adaptive Systems ◽  
Adaptive Systems ◽  
Network Models ◽  
Social Transmission ◽  
Brain Dynamics ◽  
Neural Network Models ◽  
Complex Adaptive ◽  
The Brain

Η μοντελοποίηση δυναμικών κοινωνικών διαδικασιών που λαμβάνουν χώρα στο διαδίκτυο αποτελεί ένα απαιτητικό εγχείρημα για τους παρακάτω λόγους: Πρώτον, τα πρόσωπα που αλληλεπιδρούν είναι ετερογενή και το καθένα ξεχωριστά αποτελεί ένα πολύπλοκο σύστημα. Δεύτερον, οι αλληλεπιδράσεις μεταξύ χρηστών επηρεάζονται από θόρυβο και τυχαιότητα, ενώ παράλληλα το δίκτυο των διαπροσωπικών επικοινωνιών είναι εξαιρετικά πολύπλοκο. Τρίτον, τα κοινωνικά συστήματα δεν βρίσκονται σε κατάσταση ισορροπίας καθώς η δυναμική τους επηρεάζεται από εξωτερικές διαταραχές των οποίων η κατανομή, συσχέτιση με ένα κοινωνικό σύστημα, καθώς και η μη στασιμότητά τους, είναι δύσκολο να καθοριστούν και να συμπεριληφθούν σ’ ένα δυναμικό κοινωνικό μοντέλο. Η επιτυχής μοντελοποίηση της online κοινωνικής μετάδοσης απαιτεί μια προσέγγιση ικανή να ανταπεξέλθει στις παραπάνω προκλήσεις. Γι’ αυτό το σκοπό αναπτύσσω και εφαρμόζω ένα πλαίσιο υλοποίησης βασισμένο στην θεωρία των πολύπλοκων προσαρμοστικών συστημάτων}. Μέσω μια τέτοιας μεθοδολογίας μπορεί να μελετηθεί η δυναμική φύση των αλληλεπιδράσεων μεταξύ χρηστών καθώς και οι μακροσκοπικές ιδιότητες της δραστηριότητας τους υπό την παρουσία εξωτερικών επιρροών. Ένα εξαιρετικά σύνθετο πολύπλοκο προσαρμοστικό σύστημα είναι αυτό του ανθρώπινου εγκεφάλου. Τα κοινωνικά δίκτυα είναι ακόμα πιο πολύπλοκα καθώς ουσιαστικά αποτελούνται από διασυνδεδεμένους εγκεφάλους. Ως αποτέλεσμα η μοντελοποίηση δυναμικών διαδικασιών που λαμβάνουν χώρα στα online κοινωνικά δίκτυα αποτελεί ένα υπερβολικά περίπλοκο έργο. Για την αντιμετώπιση των προκλήσεων μιας τέτοιας προσπάθειας εξετάζω τις online κοινωνικές διεργασίες μέσα από την προοπτική της νευροεπιστήμης θεωρώντας τη δυναμική των online κοινωνικών δικτύων ανάλογη με την δυναμική δικτύων αποτελούμενων από νευρώνες ολοκλήρωσης και πυροδότησης. Μέσω αυτού του ισομορφισμού εισάγω ένα νέο μοντέλο για την online κοινωνική μετάδοση το οποίο βασίζεται σε τρεις πηγές θετικής ή αρνητικής επιρροής: Την αυτοπαραγόμενη, τη διαπροσωπική και την εξωτερική. Το προτεινόμενο μοντέλο εξηγεί την ανάπτυξη της online δραστηριότητας καθώς και τις μορφές μετάδοσής της σε συνάρτηση με το δίκτυο των αλληλεπιδράσεων, την ενδογενή και εξωγενή επιρροή καθώς και τον μηχανισμό ενεργοποίησης των χρηστών. Μέσω πειραμάτων εξομοίωσης και ελέγχων εγκυρότητας των παραγόμενων αποτελεσμάτων μετά από σύγκριση με πραγματικά δεδομένα από το κοινωνικό δίκτυο Twitter, δείχνω ότι το μοντέλο αναπαράγει με ακρίβεια πρότυπα συλλογικής δραστηριότητας προερχόμενα από την απόκριση των χρηστών σε διαφόρων μορφών ερεθίσματα. Η συγκριτική αξιολόγηση των επιδόσεων του προτεινόμενου μοντέλου σε συνάρτηση με αυτή μοντέλων αναφοράς δείχνει ότι αυτό υπερτερεί σημαντικά στην ακρίβεια αναπαραγωγής πραγματικών προτύπων online δραστηριότητας. Μια ακόμα διαδικασία online κοινωνικής μετάδοσης την οποία μοντελοποιώ με καινοτόμο τρόπο σε αυτή τη Διδακτορική Διατριβή αφορά στη μετάδοση online πληροφορίας. Τη δυναμική αυτής της διαδικασίας την αναπαράγω μέσω ενός δικτυακού δυναμικού συστήματος αποτελούμενο από νευρώνες ολοκλήρωσης και πυροδότησης με θορυβώδη εισροή πληροφορίας. Μέσω του συνδυασμού ντετερμινιστικών και στοχαστικών συνιστωσών το προτεινόμενο μοντέλο αναπαράγει με ακρίβεια τα πρότυπα μετάδοσης online πληροφορίας, υποδηλώνοντας ότι αυτά εξαρτώνται από την χρονική δομή, ισχύ, καθώς και το λόγο σήματος-θορύβου των ερεθισμάτων που επηρεάζουν τους διασυνδεδεμένους χρήστες. Ο προτεινόμενος μηχανισμός ενσωματώνει τις έννοιες της ``απλής'' και ``πολύπλοκης'' μετάδοσης και επεκτείνει τις υπάρχουσες προσεγγίσεις καθώς συμπεριλαμβάνει σε ένα ενιαίο μοντέλο ενδογενείς και εξωγενείς, θετικές και αρνητικές, ντετερμινιστικές και στοχαστικές πηγές επιρροής. Τα προτεινόμενα μοντέλα νευρωνικών δικτύων είναι εύκολα προσαρμόσιμα και κατάλληλα για τη μελέτη ενός μεγάλου αριθμού από online και offline κοινωνικές δυναμικές διαδικασίες που αφορούν στη διάδοση συμπεριφορών, τάσεων και φημών, καθώς και στη διάχυση και προώθηση νέων προϊόντων. Τελικά, τη γνώση που προέκυψε από την μοντελοποίηση των προτύπων της online κοινωνικής δραστηριότητας την αξιοποιώ περαιτέρω με την ανάπτυξη ενός προβλεπτικού μοντέλου ικανού να παράγει ακριβείς προβλέψεις σχετικά με τη διάδοση online περιεχομένου.

scholarly journals Sequence structure organizes items in varied latent states of working memory neural network

eLife ◽  
2021 ◽  
Vol 10 ◽  
Author(s):  
Qiaoli Huang ◽  
Huihui Zhang ◽  
Huan Luo
Keyword(s):  
Neural Network ◽  
Working Memory ◽  
Computational Modeling ◽  
Organization Structure ◽  
Recency Effect ◽  
Sequence Structure ◽  
Sequence Context ◽  
Task Requirements ◽  
Context Shapes ◽  
The Brain

In memory experiences, events do not exist independently but are linked with each other via structure-based organization. Structure context largely influences memory behavior, but how it is implemented in the brain remains unknown. Here, we combined magnetoencephalogram (MEG) recordings, computational modeling, and impulse-response approaches to probe the latent states when subjects held a list of items in working memory (WM). We demonstrate that sequence context reorganizes WM items into distinct latent states, i.e., being reactivated at different latencies during WM retention, and the reactivation profiles further correlate with recency behavior. In contrast, memorizing the same list of items without sequence task requirements weakens the recency effect and elicits comparable neural reactivations. Computational modeling further reveals a dominant function of sequence context, instead of passive memory decaying, in characterizing recency effect. Taken together, sequence structure context shapes the way WM items are stored in the human brain and essentially influences memory behavior.

scholarly journals Task-driven hierarchical deep neural network models of the proprioceptive pathway

2020 ◽  
Author(s):  
Kai J. Sandbrink ◽  
Pranav Mamidanna ◽  
Claudio Michaelis ◽  
Mackenzie Weygandt Mathis ◽  
Matthias Bethge ◽  
...  
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Somatosensory Cortex ◽  
Action Recognition ◽  
Large Scale ◽  
Recognition Task ◽  
List Type ◽  
Control Mechanisms ◽  
Neural Network Models ◽  
3D Space

Biological motor control is versatile and efficient. Muscles are flexible and undergo continuous changes requiring distributed adaptive control mechanisms. How proprioception solves this problem in the brain is unknown. Here we pursue a task-driven modeling approach that has provided important insights into other sensory systems. However, unlike for vision and audition where large annotated datasets of raw images or sound are readily available, data of relevant proprioceptive stimuli are not. We generated a large-scale dataset of human arm trajectories as the hand is tracing the alphabet in 3D space, then using a musculoskeletal model derived the spindle firing rates during these movements. We propose an action recognition task that allows training of hierarchical models to classify the character identity from the spindle firing patterns. Artificial neural networks could robustly solve this task, and the networks’ units show directional movement tuning akin to neurons in the primate somatosensory cortex. The same architectures with random weights also show similar kinematic feature tuning but do not reproduce the diversity of preferred directional tuning nor do they have invariant tuning across 3D space. Taken together our model is the first to link tuning properties in the proprioceptive system to the behavioral level.HighlightsWe provide a normative approach to derive neural tuning of proprioceptive features from behaviorally-defined objectives.We propose a method for creating a scalable muscle spindles dataset based on kinematic data and define an action recognition task as a benchmark.Hierarchical neural networks solve the recognition task from muscle spindle inputs.Individual neural network units in middle layers resemble neurons in primate somatosensory cortex & make predictions for neurons along the proprioceptive pathway.

Sign in / Sign up

Export Citation Format

Share Document