scholarly journals Hippocampal ensembles represent sequential relationships among discrete nonspatial events

2019 ◽  
Author(s):  
Babak Shahbaba ◽  
Lingge Li ◽  
Forest Agostinelli ◽  
Mansi Saraf ◽  
Gabriel A. Elias ◽  
...  
Keyword(s):  
Spatial Information ◽  
Temporal Organization ◽  
Sequential Organization ◽  
Learning Techniques ◽  
Hippocampal Activity ◽  
Neuronal Mechanisms ◽  
Future Events ◽  
Stimulus Identity ◽  
Hippocampal Network ◽  
Sequential Firing

ABSTRACTThe hippocampus is critical to the temporal organization of our experiences, including the ability to remember past event sequences and predict future ones. Although this fundamental capacity is conserved across modalities and species, its underlying neuronal mechanisms remain poorly understood. Here we recorded hippocampal ensemble activity as rats remembered a sequence of nonspatial events (5 odor presentations unfolding over several seconds), using a task with established parallels in humans. Using novel statistical methods and deep learning techniques, we then identified new forms of sequential organization in hippocampal activity linked with task performance. We discovered that sequential firing fields (“time cells”) provided temporal information within and across events in the sequence, and that distinct types of task-critical information (stimulus identity, temporal order, and trial outcome) were also sequentially differentiated within event presentations. Finally, as previously only observed with spatial information, we report that the representations of past, present and future events were sequentially activated within individual event presentations, and that these sequential representations could be compressed within an individual theta cycle. These findings strongly suggest that a fundamental function of the hippocampal network is to encode and preserve the sequential order of experiences, and use these representations to generate predictions to inform decision-making.

Machine Learning Based Assembly of Fragments of Ancient Papyrus

2021 ◽  
Vol 14 (3) ◽  
pp. 1-21
Author(s):  
Roy Abitbol ◽  
Ilan Shimshoni ◽  
Jonathan Ben-Dov
Keyword(s):  
Machine Learning ◽  
Spatial Information ◽  
Real Life ◽  
Dead Sea ◽  
Machine Learning Techniques ◽  
Automated Classification ◽  
Learning Techniques ◽  
Test Batch ◽  
Validation Set ◽  
Local Edge

The task of assembling fragments in a puzzle-like manner into a composite picture plays a significant role in the field of archaeology as it supports researchers in their attempt to reconstruct historic artifacts. In this article, we propose a method for matching and assembling pairs of ancient papyrus fragments containing mostly unknown scriptures. Papyrus paper is manufactured from papyrus plants and therefore portrays typical thread patterns resulting from the plant’s stems. The proposed algorithm is founded on the hypothesis that these thread patterns contain unique local attributes such that nearby fragments show similar patterns reflecting the continuations of the threads. We posit that these patterns can be exploited using image processing and machine learning techniques to identify matching fragments. The algorithm and system which we present support the quick and automated classification of matching pairs of papyrus fragments as well as the geometric alignment of the pairs against each other. The algorithm consists of a series of steps and is based on deep-learning and machine learning methods. The first step is to deconstruct the problem of matching fragments into a smaller problem of finding thread continuation matches in local edge areas (squares) between pairs of fragments. This phase is solved using a convolutional neural network ingesting raw images of the edge areas and producing local matching scores. The result of this stage yields very high recall but low precision. Thus, we utilize these scores in order to conclude about the matching of entire fragments pairs by establishing an elaborate voting mechanism. We enhance this voting with geometric alignment techniques from which we extract additional spatial information. Eventually, we feed all the data collected from these steps into a Random Forest classifier in order to produce a higher order classifier capable of predicting whether a pair of fragments is a match. Our algorithm was trained on a batch of fragments which was excavated from the Dead Sea caves and is dated circa the 1st century BCE. The algorithm shows excellent results on a validation set which is of a similar origin and conditions. We then tried to run the algorithm against a real-life set of fragments for which we have no prior knowledge or labeling of matches. This test batch is considered extremely challenging due to its poor condition and the small size of its fragments. Evidently, numerous researchers have tried seeking matches within this batch with very little success. Our algorithm performance on this batch was sub-optimal, returning a relatively large ratio of false positives. However, the algorithm was quite useful by eliminating 98% of the possible matches thus reducing the amount of work needed for manual inspection. Indeed, experts that reviewed the results have identified some positive matches as potentially true and referred them for further investigation.

scholarly journals Deep learning para la clasificación de usos de suelo agrícola con Sentinel-2

2020 ◽  
pp. 35
Author(s):  
M. Campos-Taberner ◽  
F.J. García-Haro ◽  
B. Martínez ◽  
M.A. Gilabert
Keyword(s):  
Remote Sensing ◽  
Deep Learning ◽  
Near Infrared ◽  
Spatial Information ◽  
Learning Algorithm ◽  
Support Vector ◽  
Sensing Applications ◽  
Learning Techniques ◽  
Remote Sensing Applications ◽  
Sentinel 2

<p class="p1">The use of deep learning techniques for remote sensing applications has recently increased. These algorithms have proven to be successful in estimation of parameters and classification of images. However, little effort has been made to make them understandable, leading to their implementation as “black boxes”. This work aims to evaluate the performance and clarify the operation of a deep learning algorithm, based on a bi-directional recurrent network of long short-term memory (2-BiLSTM). The land use classification in the Valencian Community based on Sentinel-2 image time series in the framework of the common agricultural policy (CAP) is used as an example. It is verified that the accuracy of the deep learning techniques is superior (98.6 % overall success) to that other algorithms such as decision trees (DT), k-nearest neighbors (k-NN), neural networks (NN), support vector machines (SVM) and random forests (RF). The performance of the classifier has been studied as a function of time and of the predictors used. It is concluded that, in the study area, the most relevant information used by the network in the classification are the images corresponding to summer and the spectral and spatial information derived from the red and near infrared bands. These results open the door to new studies in the field of the explainable deep learning in remote sensing applications.</p>

scholarly journals Scopolamine Impairs Spatial Information Recorded With “Miniscope” Calcium Imaging in Hippocampal Place Cells

2021 ◽  
Vol 15 ◽  
Author(s):  
Dechuan Sun ◽  
Ranjith Rajasekharan Unnithan ◽  
Chris French
Keyword(s):  
Spatial Memory ◽  
Calcium Imaging ◽  
Memory Retrieval ◽  
Spatial Information ◽  
Place Cells ◽  
Network Activity ◽  
Memory Encoding ◽  
Neural Firing ◽  
Neural Ensemble ◽  
Hippocampal Network

The hippocampus and associated cholinergic inputs have important roles in spatial memory in rodents. Muscarinic acetylcholine receptors (mAChRs) are involved in the communication of cholinergic signals and regulate spatial memory. They have been found to impact the memory encoding process, but the effect on memory retrieval is controversial. Previous studies report that scopolamine (a non-selective antagonist of mAChR) induces cognitive deficits on animals, resulting in impaired memory encoding, but the effect on memory retrieval is less certain. We tested the effects of blocking mAChRs on hippocampal network activity and neural ensembles that had previously encoded spatial information. The activity of hundreds of neurons in mouse hippocampal CA1 was recorded using calcium imaging with a miniaturised fluorescent microscope and properties of place cells and neuronal ensemble behaviour in a linear track environment were observed. We found that the decoding accuracy and the stability of spatial representation revealed by hippocampal neural ensemble were significantly reduced after the administration of scopolamine. Several other parameters, including neural firing rate, total number of active neurons, place cell number and spatial information content were affected. Similar results were also observed in a simulated hippocampal network model. This study enhances the understanding of the function of mAChRs on spatial memory impairment.

scholarly journals Distance Transform-Based Spectral-Spatial Feature Vector for Hyperspectral Image Classification with Stacked Autoencoder

2021 ◽  
Vol 13 (9) ◽  
pp. 1732
Author(s):  
Hadis Madani ◽  
Kenneth McIsaac
Keyword(s):  
Deep Learning ◽  
Spatial Information ◽  
Hyperspectral Image ◽  
Feature Vector ◽  
Remote Sensing Data ◽  
Distance Transform ◽  
Hyperspectral Image Classification ◽  
Spatial Feature ◽  
Learning Techniques

Pixel-wise classification of hyperspectral images (HSIs) from remote sensing data is a common approach for extracting information about scenes. In recent years, approaches based on deep learning techniques have gained wide applicability. An HSI dataset can be viewed either as a collection of images, each one captured at a different wavelength, or as a collection of spectra, each one associated with a specific point (pixel). Enhanced classification accuracy is enabled if the spectral and spatial information are combined in the input vector. This allows simultaneous classification according to spectral type but also according to geometric relationships. In this study, we proposed a novel spatial feature vector which improves accuracies in pixel-wise classification. Our proposed feature vector is based on the distance transform of the pixels with respect to the dominant edges in the input HSI. In other words, we allow the location of pixels within geometric subdivisions of the dataset to modify the contribution of each pixel to the spatial feature vector. Moreover, we used the extended multi attribute profile (EMAP) features to add more geometric features to the proposed spatial feature vector. We have performed experiments with three hyperspectral datasets. In addition to the Salinas and University of Pavia datasets, which are commonly used in HSI research, we include samples from our Surrey BC dataset. Our proposed method results compares favorably to traditional algorithms as well as to some recently published deep learning-based algorithms.

scholarly journals Maintaining Consistency of Spatial Information in the Hippocampal Network: A Combinatorial Geometry Model

2016 ◽  
Vol 28 (6) ◽  
pp. 1051-1071 ◽  
Author(s):  
Y. Dabaghian
Keyword(s):  
Spatial Information ◽  
Synaptic Connection ◽  
Cell Activity ◽  
Internal Representation ◽  
Combinatorial Geometry ◽  
Place Cells ◽  
Geometry Model ◽  
Electrophysiological Recordings ◽  
Hippocampal Network ◽  
Proprioceptive Inputs

Place cells in the rat hippocampus play a key role in creating the animal’s internal representation of the world. During active navigation, these cells spike only in discrete locations, together encoding a map of the environment. Electrophysiological recordings have shown that the animal can revisit this map mentally during both sleep and awake states, reactivating the place cells that fired during its exploration in the same sequence in which they were originally activated. Although consistency of place cell activity during active navigation is arguably enforced by sensory and proprioceptive inputs, it remains unclear how a consistent representation of space can be maintained during spontaneous replay. We propose a model that can account for this phenomenon and suggest that a spatially consistent replay requires a number of constraints on the hippocampal network that affect its synaptic architecture and the statistics of synaptic connection strengths.

scholarly journals Distinct hippocampal network states support theta phase precession and theta sequences in CA1

2021 ◽  
Author(s):  
Matteo Guardamagna ◽  
Federico Stella ◽  
Francesco P. Battaglia
Keyword(s):  
Spatial Information ◽  
Place Cells ◽  
Temporal Coding ◽  
Temporal Code ◽  
Network State ◽  
Phase Precession ◽  
Hippocampal Network ◽  
Theta Phase ◽  
Network States ◽  
Theta Phase Precession

The hippocampus likely uses temporal coding to represent complex memories via mechanisms such as theta phase precession and theta sequences. Theta sequences are rapid sweeps of spikes from multiple place cells, encoding past or planned trajectories or non-spatial information. Phase precession, the correlation between a place cell's theta firing phase and animal position has been suggested to facilitate sequence emergence. We find that CA1 phase precession varies strongly across cells and environmental contingencies. Phase precession depends on the CA1 network state, and is only present when the medium gamma oscillation (60-90 Hz, linked to Entorhinal inputs) dominates. Conversely, theta sequences are most evident for non-precessing cells or with leading slow gamma (20-45 Hz, linked to CA3 inputs). These results challenge the view that phase precession is the mechanism underlying the emergence of theta sequences and point at a 'dual network states' model for hippocampal temporal code, potentially supporting merging of memory and exogenous information in CA1.

scholarly journals Relational processing demands and the role of spatial context in the construction of episodic simulations

2019 ◽  
Author(s):  
Kristina Wiebels ◽  
Donna Rose Addis ◽  
David Moreau ◽  
Valerie van Mulukom ◽  
Kelsey Esmé Onderdijk ◽  
...  
Keyword(s):  
Spatial Information ◽  
Spatial Context ◽  
Relational Processing ◽  
Theoretical Frameworks ◽  
Episodic Simulation ◽  
Processing Demands ◽  
Future Events ◽  
Event Representations ◽  
First Time

Reports on differences between remembering the past and imagining the future have led to the hypothesis that constructing future events is a more cognitively demanding process. However, factors that influence these increased demands, such as whether the event has been previously constructed and the types of details comprising the event, have remained relatively unexplored. Across two experiments, we examined how these factors influence the process of constructing event representations by having participants repeatedly construct events and measuring how construction times and a range of phenomenological ratings changed across time points. In Experiment 1, we contrasted the construction of past and future events and found that, relative to past events, the constructive demands associated with future events are particularly heightened when these events are imagined for the first time. Across repeated simulations, future events became increasingly similar to past events in terms of construction times and incorporated detail. In Experiment 2, participants imagined future events involving two memory details (person, location) and then reimagined the event either i) exactly the same, ii) with a different person, or iii) in a different location. We predicted that if generating spatial information is particularly important for event construction, a change in location will have the greatest impact on constructive demands. Results showed that spatial context contributed to these heightened constructive demands more so than person details, consistent with theories highlighting the central role of spatial processing in episodic simulation. We discuss the findings from both studies in the light of relational processing demands and consider implications for current theoretical frameworks.

scholarly journals Evidence for immediate enhancement of hippocampal memory encoding by network-targeted theta-burst stimulation during concurrent fMRI

2020 ◽  
Author(s):  
Molly S. Hermiller ◽  
Yu Fen Chen ◽  
Todd B. Parrish ◽  
Joel L. Voss
Keyword(s):  
Episodic Memory ◽  
Neural Activity ◽  
Theta Band ◽  
Theta Burst Stimulation ◽  
Burst Stimulation ◽  
Related Activity ◽  
Memory Processing ◽  
Hippocampal Activity ◽  
Hippocampal Network ◽  
Theta Burst

AbstractThe hippocampus supports episodic memory via interaction with a distributed brain network. Previous experiments using network-targeted noninvasive brain stimulation have identified episodic memory enhancements and modulation of activity within the hippocampal network. However, mechanistic insights were limited because these effects were measured long after stimulation and therefore could have reflected various neuroplastic aftereffects with extended timecourses. In this experiment with human subjects of both sexes, we tested for immediate stimulation impact on encoding-related activity of the hippocampus and immediately adjacent medial-temporal cortex by delivering theta-burst transcranial magnetic stimulation (TBS) concurrent with fMRI, as an immediate impact of stimulation would suggest an influence on neural activity. We reasoned that TBS would be particularly effective for influencing the hippocampus because rhythmic neural activity in the theta band is associated with hippocampal memory processing. First, we demonstrated that it is possible to obtain robust fMRI correlates of task-related activity during concurrent TBS. We then identified immediate effects of TBS on encoding of visual scenes. Brief volleys of TBS targeting the hippocampal network increased activity of the targeted (left) hippocampus during scene encoding and increased subsequent recollection. Stimulation did not influence activity during an intermixed numerical task with no memory demand. Control conditions using beta-band and out-of-network stimulation also did not influence hippocampal activity or recollection. TBS targeting the hippocampal network therefore immediately impacted hippocampal memory processing. This suggests direct, beneficial influence of stimulation on hippocampal neural activity related to memory and supports the role of theta-band activity in human episodic memory.Significance StatementCan noninvasive stimulation directly impact function of indirect, deep-brain targets such as the hippocampus? We tested this by targeting an accessible region of the hippocampal network via transcranial magnetic stimulation during concurrent fMRI. We reasoned that theta-burst stimulation would be particularly effective for impacting hippocampal function, as this stimulation rhythm should resonate with the endogenous theta-nested-gamma activity prominent in hippocampus. Indeed, theta-burst stimulation targeting the hippocampal network immediately impacted hippocampal activity during encoding, improving memory formation as indicated by enhanced later recollection. Rhythm- and location-control stimulation conditions had no such effects. These findings suggest a direct influence of noninvasive stimulation on hippocampal neural activity and highlight that the theta-burst rhythm is relatively privileged in its ability to influence hippocampal memory function.

scholarly journals Entrainment and maintenance of an internal metronome in premotor cortex

2018 ◽  
Author(s):  
J Cadena-Valencia ◽  
O García-Garibay ◽  
H Merchant ◽  
M Jazayeri ◽  
V de Lafuente
Keyword(s):  
Neural Activity ◽  
Nonhuman Primates ◽  
Supplementary Motor Area ◽  
Premotor Cortex ◽  
Behavioral Responses ◽  
Elapsed Time ◽  
Neuronal Mechanisms ◽  
Future Events ◽  
Motor Actions ◽  
Premotor Areas

AbstractTo prepare timely motor actions we constantly predict future events. Regularly repeating events are often perceived as a rhythm to which we can readily synchronize our movements, just as in dancing to music. However, the neuronal mechanisms underlying the capacity to encode and maintain rhythms are not understood. We trained nonhuman primates to maintain the rhythm of a visual metronome of different tempos and then we recorded neural activity in the supplementary motor area (SMA). SMA exhibited rhythmic bursts of gamma band (30-40 Hz) reflecting an internal tempo that matched the extinguished visual metronome. Moreover, gamma amplitude increased throughout the trial and provided an estimate of total elapsed time. Notably, the timing and amplitude of gamma bursts reflected systematic timing biases and errors in the behavioral responses. Our results indicate that premotor areas use dynamic motor plans to encode a metronome for rhythms and a stopwatch for total elapsed time.

scholarly journals Analysis of Bitcoin Prices using Deep Learning Techniques

2021 ◽  
Vol 12 (6) ◽  
pp. 23-27
Author(s):  
P Ravindhar Reddy Et.al
Keyword(s):  
Practical Skills ◽  
Learning Methods ◽  
Computing Power ◽  
Digital Currency ◽  
Machine Learning Methods ◽  
Learning Techniques ◽  
Conventional Technology ◽  
Future Events ◽  
Technical Material ◽  
Major Project

Project training - these projects are done for specific subjects without discussing methods that require knowledge and level of technical material. This paper discusses the design and implementation of project training in information technology as a major project that requires student creativity and describes the project clearly and concisely. These projects not only develop basic or technical skills but also practical skills. The goal of the modern student project is to show how trained machine models can predict cryptocurrency security if we allow appropriate levels of information and computing power. The diagram shows the approximate value. Conventional technology is way out of technology that can help people predict future events. With so much data being generated and stored daily, we are finally approaching an age where predictions are made from accurate and accurate data. Besides, with the advent of the digital currency era, more and more leaders have entered the digital investment market. This allows us to create models that can predict cryptocurrencies, especially Bitcoin. This can be done through several types of machine learning methods and in-depth learning and learning methods.

Sign in / Sign up

Export Citation Format

Share Document