(Invited) Regression and Optimization by Analogue In-memory Computing in Crosspoint Arrays

Constraint satisfaction and search problems generally fall into the class of problems for which a direct algorithmic solution does not exist. The solution of these problems requires the examination of state spaces. A problem solver (inference engine) alone is not able to organize and maintain state space consistently. For this purpose a so‐called truth maintenance system is required. Our truth maintenance system (MEKON) organizes data within a data abstraction called a context. Each context corresponds to one problem state and contains currently believed data. The truth maintenance system provides believed data retrieval, belief revision, contradiction handling and non‐monotonicity handling, the features that help a problem solver to examine state spaces. MEKON represents an ATMS‐like system implemented within BEST (Blackboard‐based Expert System Toolkit). However, some special MEKON features such as state variables, context sensitive generation of assumptions and explicit context generation, that are not present in standard ATMSs, facilitate not only the solution of constraint satisfaction problems, but also the solution of search problems (not provided by standard ATMSs). Being deeply integrated with Prolog/Rex, BEST’s knowledge representation language, and BEST’s inference engine, MEKON provides a simple and efficient means for the examination of state spaces. Facts, hypotheses (assumptions), and concepts (frames) are used to describe a problem state, contexts are used to represent points in the state spaces, while rules are used to perform state transitions. MEKON is the only truth maintenacne system that provides truth maintenance capabilities on local blackboards thus enabling the solution of complex problems which include different kinds of real constraint satisfaction and search problems concerning diagnosis, allocation tasks, classification tasks, planning or scenario making.

The bakery sector is experiencing growth, with enterprises like XY Bakery & Cake Shop providing their products directly to consumers. The company's bread sales have exhibited irregular pattern, rendering the task of establishing precise sales forecasts difficult. This inconsistency has resulted in financial losses, such as the spoilage of bread within three days, the squandering of raw materials, labor, and other forms of inefficiency. The objective of this study is to forecast bread sales and evaluate the accuracy of the predictions using the Extreme Learning Machine (ELM) approach. The accuracy will be assessed using the Mean Square Error (MSE) measure at AR Bakery. The ELM approach is utilized to predict bread sales, using normalized data divided into 80% for training and 20% for testing. The experimental tests conducted with 8 neurons produced mean squared error (MSE) value of 0.27402, whereas utilizing 5 neurons resulted in an MSE of 0.28761. The minimum error value was achieved using 5 neurons. 10 predictions were made using a dataset consisting of 101 data points for each type of bread. In December, the sales reached their peak, with a total of 198,850 units sold.

Matrix-vector multiplication (MVM) is the core operation of many important algorithms. Crosspoint resistive memory array enables naturally calculating MVM in one operation, thus representing a highly promising computing accelerator for various applications. To evaluate computing performance as well as scalability of in-memory MVM, the fundamental issue of time complexity of the circuit shall be elaborated. Based on the most common MVM circuit that uses transimpedance amplifiers to read out current product in crosspoint array, we analyze its dynamic response and the corresponding time complexity. The result shows that the computing time is governed by the maximal row sum of the implemented matrix, which leads to an explicit time complexity for a specific dataset, e.g., O( N <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1/2</sup> ) and O(ln N) for discrete cosine transformation and Toeplitz matrix, respectively. By changing accordingly feedback conductance of transimpedance amplifier for different matrix sizes, it is possible to reduce the time complexity to O(1). Impact of non-ideal factors of the circuit on computing time is also studied. This work provides an insight into the performance and its improvement of MVM computation for efficient in-memory computing accelerators.

ENIC: Ensemble and Nature Inclined Classification with Sparse Depiction based Deep and Transfer Learning for Biosignal Classification

On extreme learning machines in sequential and time series prediction: A non-iterative and approximate training algorithm for recurrent neural networks

Multi-core processors have been broadly available to the public in the last five years. Parallelism has become a common design feature for computational intensive algorithms. In this paper we present a parallel implementation of an algorithm called interval constraint propagation for solution of constraint satisfaction problems over real numbers. Unlike existing implementations of this algorithm, our implementation scales well to many CPU cores with shared memory for sparse constraint satisfaction problems. We present scalability data for a quad-core processor on a number of benchmarks for non-linear constraint solvers.

An approach1 is proposed to organize the local search procedures in the constraint satisfaction problems, which is based on a "compressed" representation of qualitative constraints in the form of matrix-like structures. Formulated are the theorems which allow, on the basis of analysis of one particular solution and internal structure of non-numerical constraints matrices, revelation of subspaces in spaces of legal /or illegal assignments, which contributes to speed-up computational procedures.

Parsing is a key task in computer science, with applications in compilers, natural language processing, syntactic pattern matching, and formal language theory. With the recent development of deep learning techniques, several artificial intelligence applications, especially in natural language processing, have combined traditional parsing methods with neural networks to drive the search in the parsing space, resulting in hybrid architectures using both symbolic and distributed representations. In this article, we show that existing symbolic parsing algorithms for context-free languages can cross the border and be entirely formulated over distributed representations. To this end, we introduce a version of the traditional Cocke–Younger–Kasami (CYK) algorithm, called distributed (D)-CYK, which is entirely defined over distributed representations. D-CYK uses matrix multiplication on real number matrices of a size independent of the length of the input string. These operations are compatible with recurrent neural networks. Preliminary experiments show that D-CYK approximates the original CYK algorithm. By showing that CYK can be entirely performed on distributed representations, we open the way to the definition of recurrent layer neural networks that can process general context-free languages.

Recently, there has been an ever-increasing interest in the study of data-driven analytics to predict hidden dangers in the cloud service-based coal mine production, with the purpose of the prevention of possible accidents. In this paper, to achieve the above prediction, a machine learning algorithm based on the single-hidden layer feed forward network (SLFN) using timeliness managing extreme learning machine (TMELM) is utilized. Compared with those traditional learning algorithms, extreme learning machine (ELM) has its unique feature of a higher generalization capability at a much faster learning speed. In addition, the timeliness managing ELM has been proposed by incorporating timeliness management scheme into ELM approach. Under the timeliness managing ELM scheme used to predict the hidden dangers, the newly incremental data could be prior to the historical data while maximizing the contribution of the newly increasing training data, since it may be more feasible that the incremental data can contribute reasonable weights to represent the current production situation in accordance with the practical analysis for accidents in coal mine production. The experimental results on the coal mines of Beijing show that by using timeliness managing ELM, the prediction accuracy of hidden danger can be improved with better stability compared with other similar machine learning methods.

An improved classification approach is proposed to solve the hot research problem of some complex multiclassification samples based on extreme learning machine (ELM). ELM was proposed based on the single‐hidden layer feed‐forward neural network (SLFNN). ELM is characterized by the easier parameter selection rules, the faster converge speed, the less human intervention, and so on. In order to further improve the classification precision of ELM, an improved generation method of the network structure of ELM is developed by dynamically adjusting the number of hidden nodes. The number change of the hidden nodes can serve as the computational updated step length of the ELM algorithm. In this paper, the improved algorithm can be called the variable step incremental extreme learning machine (VSI‐ELM). In order to verify the effect of the hidden layer nodes on the performance of ELM, an open‐source machine learning database (University of California, Irvine (UCI)) is provided by the performance test data sets. The regression and classification experiments are used to study the performance of the VSI‐ELM model, respectively. The experimental results show that the VSI‐ELM algorithm is valid. The classification of different degrees of broken wires is now still a problem in the nondestructive testing of hoisting wire rope. The magnetic flux leakage (MFL) method of wire rope is an efficient nondestructive method which plays an important role in safety evaluation. Identifying the proposed VSI‐ELM model is effective and reliable for actually applying data, and it is used to identify the classification problem of different types of samples from MFL signals. The final experimental results show that the VSI‐ELM algorithm is of faster classification speed and higher classification accuracy of different broken wires.

Surface roughness prediction by extreme learning machine constructed with abrasive water jet

This paper proposes a procedure to estimate tsunami wave forces on coastal bridges through a novel method based on Extreme Learning Machine (ELM) and laboratory experiments. This research included three water depths, ten wave heights, and four bridge models with a variety of girders providing a total of 120 cases. The research was designed and adapted to estimate tsunami bore forces including horizontal force, vertical uplift and overturning moment on a coastal bridge. The experiments were carried out on 1:40 scaled concrete bridge models in a wave flume with dimensions of 24 m × 1.5 m × 2 m. Two six-axis load cells and four pressure sensors were installed to the base plate to measure forces. In the numerical procedure, estimation and prediction results of the ELM model were compared with Genetic Programming (GP) and Artificial Neural Networks (ANNs) models. The experimental results showed an improvement in predictive accuracy, and capability of generalization could be achieved by the ELM approach in comparison with GP and ANN. Moreover, results indicated that the ELM models developed could be used with confidence for further work on formulating novel model predictive strategy for tsunami bore forces on a coastal bridge. The experimental results indicated that the new algorithm could produce good generalization performance in most cases and could learn thousands of times faster than conventional popular learning algorithms. Therefore, it can be conclusively obtained that utilization of ELM is certainly developing as an alternative approach to estimate the tsunami bore forces on a coastal bridge.

The rapid progress in artificial intelligence (AI) is driving the need for more efficient computing architectures capable of performing the demanding, repetitive computations required for processing large datasets. Unlike traditional von Neumann architectures, which separate processing and memory units, new approaches aim to integrate computational capabilities directly into memory arrays, facilitating in-memory computing (IMC). This integration significantly improves both energy efficiency and processing speed, particularly for AI and similar applications. A crucial operation within IMC is vector-matrix multiplication (VMM), where input signal vectors interact with a memory state matrix to produce output signal vectors.One innovative approach to implementing VMM is through the use of a ferroelectric Hf₀.₅Zr₀.₅O₂ (HZO) layer in conventional field-effect transistor (FET) or metal-insulator-metal (MIM) devices. These HZO-based ferroelectric components show significant promise for low-power, IMC-compatible systems, especially when integrated into MIM capacitor structures. Their simple fabrication process supports scalable system integration, while offering benefits such as low power consumption and high reliability. However, current IMC techniques utilizing ferroelectric MIM capacitors rely on small-signal capacitance readout, which suffers from limited contrast between high and low capacitance states, making it difficult to distinguish memory states effectively for efficient IMC operations.In this study, we introduce two novel readout techniques to improve the scalability and robustness of IMC, focusing on enhanced memory differentiation by reading the ferroelectric remanent polarization (Pᵣ) instead of relying on small-signal capacitance. These methods exploit the switching behavior between two stable polarization states (Pₗₒʷ and Pᵤₚ) of the HZO material. To measure charge accumulation resulting from polarization switching, we connect the ferroelectric HZO capacitors in series with either a shunt resistor (Rₛₕᵤₙₜ) or an external capacitor (Cₑₓt) to produce current or voltage signals, respectively. Both readout methods show significantly improved memory differentiation and VMM performance compared to the traditional small-signal capacitance approach. For example, in the current readout method, the maximum displacement current in the high-current state (Iₛʷ + Iₙₒₙ-ˢʷ) is approximately 0.31 mA, while the low-current state (Iₙₒₙ-ˢʷ) is about 0.04 mA when no switching current is present. Notably, the static leakage current during the input pulse is nearly zero, underscoring the potential for low-power operation in capacitive systems. In the voltage readout method, an initial Pᵤₚ state leads to a high output voltage (Vₒᵤₜ) of approximately 0.13 V, while an initial Pₗₒʷ state produces a low Vₒᵤₜ of about 0 V. These polarization-based readout techniques provide substantially higher memory distinction and better VMM performance compared to traditional small-signal capacitance methods.We will demonstrate IMC systems by using voltage readout of the Pr change in HZO layers, employing charge compensation between the Cext and HZO capacitors. This method takes advantage of the superior memory distinction within the capacitive system, offering more reliable and lower-power performance compared to traditional Fe-FET approaches.The authors acknowledge Prof. Joseph S. Friedman and Dr. Peng Zhou for the technical discussion.

The trend prediction of the stock is a main challenge. Accidental factors often lead to short-term sharp fluctuations in stock markets, deviating from the original normal trend. The short-term fluctuation of stock price has high noise, which is not conducive to the prediction of stock trends. Therefore, we used discrete wavelet transform (DWT)-based denoising to denoise stock data. Denoising the stock data assisted us to eliminate the influences of short-term random events on the continuous trend of the stock. The denoised data showed more stable trend characteristics and smoothness. Extreme learning machine (ELM) is one of the effective training algorithms for fully connected single-hidden-layer feedforward neural networks (SLFNs), which possesses the advantages of fast convergence, unique results, and it does not converge to a local minimum. Therefore, this paper proposed a combination of ELM- and DWT-based denoising to predict the trend of stocks. The proposed method was used to predict the trend of 400 stocks in China. The prediction results of the proposed method are a good proof of the efficacy of DWT-based denoising for stock trends, and showed an excellent performance compared to 12 machine learning algorithms (e.g., recurrent neural network (RNN) and long short-term memory (LSTM)).

In this paper, we propose an recurrent neural network (RNN) based step-size estimation for the recursive least squares (RLS) algorithm with application to acoustic echo cancellation (AEC). RLS-based AEC (as compared to the least mean square (LMS) based) has a better convergence rate and less distortion, which is an important advantage in acoustic speech recognition (ASR) usages. The optimal step size of the RLS, as derived in the literature, is impractical due to its dependence on the power of the desired speech. An RNN model is presented that learns the relationship between the reference signal and the microphone signal to the optimal step size. At inference mode, the trained RNN produces the step size in real-time, which is then fed to the RLS algorithm. For evaluation, we used two hours of recordings from the AEC-challenge and Wav2vec2 databases, and compared the proposed technique and other competing methods. Experiments show that the proposed technique has advantages in terms of ASR performance and other classic measurements.