Hierarchical Temporal Memory Model Research Articles

Time series processing-based malicious activity detection in SCADA systems

Many critical infrastructures, essential to modern life, such as oil and gas pipeline control and electricity distribution, are managed by SCADA systems. In the contemporary landscape, these systems are interconnected to the internet, rendering them vulnerable to numerous cyber-attacks. Consequently, ensuring SCADA security has become a crucial area of research. This paper focuses on detecting attacks that manipulate the timing of commands within the system, while maintaining their original order and content. To address this challenge, we propose several machine-learning-based methods. The first approach relies on Long-Short-Term Memory model, and the second utilizes Hierarchical Temporal Memory model, both renowned for their effectiveness in detecting patterns in time-series data. We rigorously evaluate our methods using a real-life SCADA system dataset and show that they outperform previous techniques designed to combat such attacks.

Extracting Geoscientific Dataset Names from the Literature Based on the Hierarchical Temporal Memory Model

Extracting geoscientific dataset names from the literature is crucial for building a literature–data association network, which can help readers access the data quickly through the Internet. However, the existing named-entity extraction methods have low accuracy in extracting geoscientific dataset names from unstructured text because geoscientific dataset names are a complex combination of multiple elements, such as geospatial coverage, temporal coverage, scale or resolution, theme content, and version. This paper proposes a new method based on the hierarchical temporal memory (HTM) model, a brain-inspired neural network with superior performance in high-level cognitive tasks, to accurately extract geoscientific dataset names from unstructured text. First, a word-encoding method based on the Unicode values of characters for the HTM model was proposed. Then, over 12,000 dataset names were collected from geoscience data-sharing websites and encoded into binary vectors to train the HTM model. We conceived a new classifier scheme for the HTM model that decodes the predictive vector for the encoder of the next word so that the similarity of the encoders of the predictive next word and the real next word can be computed. If the similarity is greater than a specified threshold, the real next word can be regarded as part of the name, and a successive word set forms the full geoscientific dataset name. We used the trained HTM model to extract geoscientific dataset names from 100 papers. Our method achieved an F1-score of 0.727, outperforming the GPT-4- and Claude-3-based few-shot learning (FSL) method, with F1-scores of 0.698 and 0.72, respectively.

Improved Hierarchical Temporal Memory for Online Prediction of Ocean Time Series Data

Time series prediction is an effective tool for marine scientific research. The Hierarchical Temporal Memory (HTM) model has advantages over traditional recurrent neural network (RNN)-based models due to its online learning and prediction capabilities. Given that the neuronal structure of HTM is ill-equipped for the complexity of long-term marine time series applications, this study proposes a new, improved HTM model, incorporating Gated Recurrent Units (GRUs) neurons into the temporal memory algorithm to overcome this limitation. The capacities and advantages of the proposed model were tested and evaluated on time series data collected from the Xiaoqushan Seafloor Observatory in the East China Sea. The improved HTM model both outperforms the original one in short-term and long-term predictions and presents results with lower errors and better model stability than the GRU model, which is proficient in long-term predictions. The findings allow for the conclusion that the mechanism of online learning has certain advantages in predicting ocean observation data.

NeoSLAM: Long-Term SLAM Using Computational Models of the Brain.

Simultaneous Localization and Mapping (SLAM) is a fundamental problem in the field of robotics, enabling autonomous robots to navigate and create maps of unknown environments. Nevertheless, the SLAM methods that use cameras face problems in maintaining accurate localization over extended periods across various challenging conditions and scenarios. Following advances in neuroscience, we propose NeoSLAM, a novel long-term visual SLAM, which uses computational models of the brain to deal with this problem. Inspired by the human neocortex, NeoSLAM is based on a hierarchical temporal memory model that has the potential to identify temporal sequences of spatial patterns using sparse distributed representations. Being known to have a high representational capacity and high tolerance to noise, sparse distributed representations have several properties, enabling the development of a novel neuroscience-based loop-closure detector that allows for real-time performance, especially in resource-constrained robotic systems. The proposed method has been thoroughly evaluated in terms of environmental complexity by using a wheeled robot deployed in the field and demonstrated that the accuracy of loop-closure detection was improved compared with the traditional RatSLAM system.

Multi-ant colony optimization algorithm based on game strategy and hierarchical temporal memory model

Multi-ant colony optimization algorithm based on game strategy and hierarchical temporal memory model

The Cup Whisperer

The Cup Whisperer

A New Hierarchical Temporal Memory Algorithm Based on Activation Intensity.

As a human-cortex-inspired computing model, hierarchical temporal memory (HTM) has shown great promise in sequence learning and has been applied to various time-series applications. HTM uses the combination of columns and neurons to learn the temporal patterns within the sequence. However, the conventional HTM model compacts the input into two naive column states—active and nonactive, and uses a fixed learning strategy. This simplicity limits the representation capability of HTM and ignores the impacts of active columns on learning the temporal context. To address these issues, we propose a new HTM algorithm based on activation intensity. By introducing the column activation intensity, more useful and fine-grained information from the input is retained for sequence learning. Furthermore, a self-adaptive nonlinear learning strategy is proposed where the synaptic connections are dynamically adjusted according to the activation intensity of columns. Extensive experiments are carried out on two real-world time-series datasets. Compared to the conventional HTM and LSTM model, our method achieved higher accuracy and less time overhead.

Waste-to-energy forecasting and real-time optimization: An anomaly-aware approach

Waste-to-energy (WTE) technologies convert municipal solid, and biomass wastes into affordable renewable heat and power energy. However, there are large uncertainties associated with using waste feed as a renewable energy source. This paper proposes a WTE management tool that provides forecasting and real-time optimization of power generated with the consideration of anomaly. The WTE management framework was designed based on a biological neural network, the Hierarchical Temporal Memory (HTM) coupled with a dual-mode optimization procedure. The HTM model is inspired by the mechanism in the cerebral neocortex of the brain, providing anomaly identification and spatial-temporal prediction. In this work, the HTM-based smart energy framework is demonstrated in an industrial case study for the power generation of a waste-to-energy cogeneration system. HTM was compared with methods such as Long Short-Term Memory (LSTM) neural network, Autoregressive Integrated Moving Average (ARIMA), Fourier Transformation Extrapolation (FTE), persistence forecasting, and was able to achieve mean squared error (MSE) of 0.08466% while giving 35450 Euro profit in half a year. Coupled with a novel dual-mode optimization procedure, HTM demonstrated 11% improvement with respect to only predictive optimization (with HTM) in estimated gross profit.

Anomalous Behavior Detection Framework Using HTM-Based Semantic Folding Technique

Upon the working principles of the human neocortex, the Hierarchical Temporal Memory model has been developed which is a proposed theoretical framework for sequence learning. Both categorical and numerical types of data are handled by HTM. Semantic Folding Theory (SFT) is based on HTM to represent a data stream for processing in the form of sparse distributed representation (SDR). For natural language perception and production, SFT delivers a solid structural background for semantic evidence description to the fundamentals of the semantic foundation during the phase of language learning. Anomalies are the patterns from data streams that do not follow the expected behavior. Any stream of data patterns could have a number of anomaly types. In a data stream, a single pattern or combination of closely related patterns that diverges and deviates from standard, normal, or expected is called a static (spatial) anomaly. A temporal anomaly is a set of unexpected changes between patterns. When a change first appears, this is recorded as an anomaly. If this change looks a number of times, then it is set to a “new normal” and terminated as an anomaly. An HTM system detects the anomaly, and due to continuous learning nature, it quickly learns when they become the new normal. A robust anomalous behavior detection framework using HTM-based SFT for improving decision-making (SDR-ABDF/P2) is a proposed framework or model in this research. The researcher claims that the proposed model would be able to learn the order of several variables continuously in temporal sequences by using an unsupervised learning rule.

Hierarchical Temporal Memory-Based One-Pass Learning for Real-Time Anomaly Detection and Simultaneous Data Prediction in Smart Grids

A neuro-cognitive inspired architecture based on the Hierarchical Temporal Memory (HTM) is proposed for anomaly detection and simultaneous data prediction in real-time for smart grid <inline-formula><tex-math notation="LaTeX">$\mu$</tex-math><alternatives><mml:math><mml:mi>&#x03BC;</mml:mi></mml:math><inline-graphic xlink:href="barua-ieq1-3037054.gif"/></alternatives></inline-formula>PMU data. The key technical idea is that the HTM learns a <i>sparse distributed temporal representation</i> of sequential data that turns out to be very useful for anomaly detection and simultaneous data prediction in real-time. Our results show that the proposed HTM can predict anomalies within 83&#x2013;90 percent accuracy for three different application profiles, namely <i>Standard, Reward Few False Positive, Reward Few False Negative</i> for two different datasets. We show that the HTM is competitive to five state-of-the-art algorithms for anomaly detection. Moreover, for the multi-step prediction in the online setting, the same HTM achieves a low 0.0001 normalized mean square error, a low negative log-likelihood score of 1.5 and is also competitive to six state-of-the-art prediction algorithms. We demonstrate that the same HTM model can be used for <i>both the tasks</i> and can learn online in one-pass, in an unsupervised fashion and adapt to changing statistics. <i>The other state-of-the-art algorithms are either less accurate or are limited to one of the tasks or cannot learn online in one-pass, and adapt to changing statistics.</i>

COMPARATIVE ANALYSIS OF SYSTEM LOGS AND STREAMING DATA ANOMALY DETECTION ALGORITHMS

This paper provides with the description, comparative analysis of multiple commonly used approaches of the analysis of system logs, and streaming data massively generated by company IT infrastructure with an unattended anomaly detection feature. An importance of the anomaly detection is dictated by the growing costs of system downtime due to the events that would have been predicted based on the log entries with the abnormal data reported. Anomaly detection systems are built using standard workflow of the data collection, parsing, information extraction and detection steps. Most of the document is related to the anomaly detection step and algorithms like regression, decision tree, SVM, clustering, principal components analysis, invariants mining and hierarchical temporal memory model. Model-based anomaly algorithms and hierarchical temporary memory algorithms were used to process HDFS, BGL and NAB datasets with ~16m log messages and 365k data points of the streaming data. The data was manually labeled to enable the training of the models and accuracy calculation. According to the results, supervised anomaly detection systems achieve high precision but require significant training effort, while HTM-based algorithm shows the highest detection precision with zero training. Detection of the abnormal system behavior plays an important role in large-scale incident management systems. Timely detection allows IT administrators to quickly identify issues and resolve them immediately. This approach reduces the system downtime dramatically.Most of the IT systems generate logs with the detailed information of the operations. Therefore, the logs become an ideal data source of the anomaly detection solutions. The volume of the logs makes it impossible to analyze them manually and requires automated approaches.

Multivariate-Time-Series-Driven Real-time Anomaly Detection Based on Bayesian Network.

Anomaly detection is an important research direction, which takes the real-time information system from different sensors and conditional information sources into consideration. Based on this, we can detect possible anomalies expected of the devices and components. One of the challenges is anomaly detection in multivariate-sensing time-series in this paper. Based on this situation, we propose RADM, a real-time anomaly detection algorithm based on Hierarchical Temporal Memory (HTM) and Bayesian Network (BN). First of all, we use HTM model to evaluate the real-time anomalies of each univariate-sensing time-series. Secondly, a model of anomalous state detection in multivariate-sensing time-series based on Naive Bayesian is designed to analyze the validity of the above time-series. Lastly, considering the real-time monitoring cases of the system states of terminal nodes in Cloud Platform, the effectiveness of the methodology is demonstrated using a simulated example. Extensive simulation results show that using RADM in multivariate-sensing time-series is able to detect more abnormal, and thus can remarkably improve the performance of real-time anomaly detection.

A Distributed Anomaly Detection System for In-Vehicle Network Using HTM

With the development of 5G and Internet of Vehicles technology, the possibility of remote wireless attack on an in-vehicle network has been proven by security researchers. Anomaly detection technology can effectively alleviate the security threat, as the first line of security defense. Based on this, this paper proposes a distributed anomaly detection system using hierarchical temporal memory (HTM) to enhance the security of a vehicular controller area network bus. The HTM model can predict the flow data in real time, which depends on the state of the previous learning. In addition, we improved the abnormal score mechanism to evaluate the prediction. We manually synthesized field modification and replay attack in data field. Compared with recurrent neural networks and hidden Markov model detection models, the results show that the distributed anomaly detection system based on HTM networks achieves better performance in the area under receiver operating characteristic curve score, precision, and recall.

Continuous Online Sequence Learning with an Unsupervised Neural Network Model.

The ability to recognize and predict temporal sequences of sensory inputs is vital for survival in natural environments. Based on many known properties of cortical neurons, hierarchical temporal memory (HTM) sequence memory recently has been proposed as a theoretical framework for sequence learning in the cortex. In this letter, we analyze properties of HTM sequence memory and apply it to sequence learning and prediction problems with streaming data. We show the model is able to continuously learn a large number of variable order temporal sequences using an unsupervised Hebbian-like learning rule. The sparse temporal codes formed by the model can robustly handle branching temporal sequences by maintaining multiple predictions until there is sufficient disambiguating evidence. We compare the HTM sequence memory with other sequence learning algorithms, including statistical methods-autoregressive integrated moving average; feedforward neural networks-time delay neural network and online sequential extreme learning machine; and recurrent neural networks-long short-term memory and echo-state networks on sequence prediction problems with both artificial and real-world data. The HTM model achieves comparable accuracy to other state-of-the-art algorithms. The model also exhibits properties that are critical for sequence learning, including continuous online learning, the ability to handle multiple predictions and branching sequences with high-order statistics, robustness to sensor noise and fault tolerance, and good performance without task-specific hyperparameter tuning. Therefore, the HTM sequence memory not only advances our understanding of how the brain may solve the sequence learning problem but is also applicable to real-world sequence learning problems from continuous data streams.

The biologically inspired Hierarchical Temporal Memory Model for Farsi Handwritten Digit and Letter Recognition

The biologically inspired Hierarchical Temporal Memory Model for Farsi Handwritten Digit and Letter Recognition

License plate recognition algorithm on the basis of a connected components method and a hierarchical temporal memory model

This paper proposes a license plate recognition algorithm that consists of three major steps: image preprocessing, segmentation, and recognition, which works efficiently with day- and nighttime images, as well as with the license plate being tilted. Pre-filtration allows the sequential binarization to be conducted efficiently. Typically, the license plate segmentation is realized by a histogram method with the preliminary plate de-rotation to the horizontal position, thus deteriorating the original image quality. In this paper the segmentation is implemented by a connected components method, enabling the rotation and a consequent loss of quality to be avoided. The hierarchical temporal network shows good results in rotated symbols recognition. The proposed method can be used in a similar way for segmentation and recognition of various text data. The proposed algorithms can also be used for distorted text segmentation and recognition.

Each to their own beat: periodicity in temporal inference

Characterizing cortical computation has been a key challenge in neuroscience. It has been suggested that the cerebral cortex is involved in unsupervised learning [1] and the notion that cortical computation is probabilistic is being increasingly accepted [2]. It has also been shown that cortical regions are hierarchically connected [3] and this has important implications on the way information could be represented. Moreover, the classical notion that processing in the cortex is predominantly governed by its feedforward inputs has been largely replaced by the view that feedback signals and other contextual information play a crucial role in the way information is encoded and learnt [2,4,5]. There is increasing emphasis on the influence of ongoing activity [5], ie. the state of the network on the effects of feedforward inputs. The hierarchical temporal memory (HTM) model of cortical computation [6] was used as the starting point for the current work. HTM is a hierarchical probabilistic model that uses spatial and temporal characteristics of its inputs to represent 'invariant causes' in the world. The key assumptions in the model are that learning in the cortex involves learning sequences of coincident patterns in its inputs and that the temporal patterns in inputs are critical for the brain’s ability to learn and infer with ambiguous sensory information. For the current work we are investigating tactile perception and implementing the HTM model using sensory inputs from artificial whiskers. The HTM model emphasizes that temporal patterns in inputs are extremely important features for learning and inference in the brain. This is especially the case with tactile perception since very little information is available at any single instant; how sensory inputs evolve over time conveys indispensable information about stimuli in the world. In modeling unsupervised learning, ‘time’ is commonly treated as indexing sequential inputs [7]. However, we are suggesting that learning, inference and prediction with temporal patterns pose a theoretical need to model some mechanism to encode time not only in terms of a temporal succession of events and causes, but also as a metric of periodicity. This theoretical need is perfectly aligned with well established empirical evidence that the brain engages in a plethora of oscillations at several different scales and frequencies. Our implementation of HTM incorporates a mechanism which attempts to mimic hypothesized effects of local neural oscillations in temporal inference. We are suggesting that cortical networks can dynamically maintain local oscillations to best match the relative frequencies with which inputs tend to occur. Assuming that all inputs to any cortical network are superimposed on its ongoing activity, these local oscillations can discretize the continuous stream of sensory inputs in signaling to subsequent levels in the hierarchy, as well as modulate the signal to suppress or enhance effects of specific components in the inputs based on their timing. We are demonstrating these effects in our implementation of HTM.

Object Identification in Dynamic Images Based on the Memory-Prediction Theory of Brain Function

In 2004, Jeff Hawkins presented a memory-prediction theory of brain function, and later used it to create the Hierar-chical Temporal Memory model. Several of the concepts described in the theory are applied here in a computer vision system for a mobile robot application. The aim was to produce a system enabling a mobile robot to explore its envi-ronment and recognize different types of objects without human supervision. The operator has means to assign names to the identified objects of interest. The system presented here works with time ordered sequences of images. It utilizes a tree structure of connected computational nodes similar to Hierarchical Temporal Memory and memorizes frequent sequences of events. The structure of the proposed system and the algorithms involved are explained. A brief survey of the existing algorithms applicable in the system is provided and future applications are outlined. Problems that can arise when the robot’s velocity changes are listed, and a solution is proposed. The proposed system was tested on a sequence of images recorded by two parallel cameras moving in a real world environment. Results for mono- and ste-reo vision experiments are presented.