Continual learning on edge platforms remains challenging because recurrent networks depend on energy-intensive training procedures and frequent data movement that are impractical for embedded deployments. This work introduces M2RU, a mixed-signal architecture that implements the minion recurrent unit for efficient temporal processing with on-chip continual learning. The architecture integrates weighted-bit streaming, which enables multi-bit digital inputs to be processed in crossbars without high-resolution conversion, and an experience replay mechanism that stabilizes learning under domain shifts. M2RU achieves ∼13 GOPS at 16.76 mW, corresponding to 776 GOPS per watt, and maintains accuracy within 5 percent of software baselines on sequential MNIST, CIFAR-10, and Google Speech Commands tasks. Compared with a CMOS digital design, the accelerator provides 25× improvement in energy efficiency. Device-aware analysis shows an expected operational lifetime of 12.2 years under continual learning workloads. These results establish M2RU as a scalable and energy-efficient platform for real-time adaptation in edge-level temporal intelligence.
Investigating the strength and the relationship between the Self-organized learning strategies and self-competence among talented students was the aim of this study. To do this, the researcher employed the correlation descriptive approach, whereby a sample of (120) male and female student were selected from various Iraqi cities for the academic year 2015-2016. the researcher setup two scales based on the previous studies: one to measure the Self-organized learning strategies which consist of (47) item and the other to measure the self-competence that composed of (50) item. Both of these scales were applied on the targeted sample to collect the required data
In this article, the research presents a general overview of deep learning-based AVSS (audio-visual source separation) systems. AVSS has achieved exceptional results in a number of areas, including decreasing noise levels, boosting speech recognition, and improving audio quality. The advantages and disadvantages of each deep learning model are discussed throughout the research as it reviews various current experiments on AVSS. The TCD TIMIT dataset (which contains top-notch audio and video recordings created especially for speech recognition tasks) and the Voxceleb dataset (a sizable collection of brief audio-visual clips with human speech) are just a couple of the useful datasets summarized in the paper that can be used to t
... Show MoreIn this article, the research presents a general overview of deep learning-based AVSS (audio-visual source separation) systems. AVSS has achieved exceptional results in a number of areas, including decreasing noise levels, boosting speech recognition, and improving audio quality. The advantages and disadvantages of each deep learning model are discussed throughout the research as it reviews various current experiments on AVSS. The TCD TIMIT dataset (which contains top-notch audio and video recordings created especially for speech recognition tasks) and the Voxceleb dataset (a sizable collection of brief audio-visual clips with human speech) are just a couple of the useful datasets summarized in the paper that can be used to test A
... Show MoreGiven the importance of possessing the digital competence (DC) required by the technological age, whether for teachers or students and even communities and governments, educational institutions in most countries have sought to benefit from modern technologies brought about by the technological revolution in developing learning and teaching and using modern technologies in providing educational services to learners. Since university students will have the doors to work opened in all fields, the research aims to know their level of DC in artificial intelligence (AI) applications and systems utilizing machine learning (ML) techniques. The descriptive approach was used, as the research community consisted of students from the University
... Show MoreThis paper proposes a better solution for EEG-based brain language signals classification, it is using machine learning and optimization algorithms. This project aims to replace the brain signal classification for language processing tasks by achieving the higher accuracy and speed process. Features extraction is performed using a modified Discrete Wavelet Transform (DWT) in this study which increases the capability of capturing signal characteristics appropriately by decomposing EEG signals into significant frequency components. A Gray Wolf Optimization (GWO) algorithm method is applied to improve the results and select the optimal features which achieves more accurate results by selecting impactful features with maximum relevance
... Show MoreTo observe the effect of media of the internal pressure on the equivalent stress distribution in the tube, an experimental study is done by constructing a testing rig to apply the hydraulic pressure and three dies are manufactured with different bulging configurations (square, cosine, and conical). In the other part, ANSYS APDL is generated to analyze the bulging process with hydraulic and rubber (natural and industrial) media. It was found that when the media is a rubber, the stress is decreased about 9.068% in case of cosine die and 5.4439% in case of conical die and 2.8544% in case of square die. So, it can be concluded that the internal pressure in the rubber media is much better than in hydraulic media. Also, the force needed for fo
... Show MoreSurfaces quality is one of the most specified customer requirements for machine parts. The major indication of surfaces quality on machined parts is surface roughness. The research aim is to study the cutting conditions and their effects on the surface roughness. This paper utilizes regression models to predict surface roughness over the machining time for variety of cutting conditions in turning. In the experimental part for turning, different types of materials (Aluminum alloy, Copper alloy, and Gray cast iron) were considered with different cutting speed ( ) and feed rate ( ). A mathematical Model depending on statistical-mathematical method between surface roughness (Rz ) and cutting condition ( , ) were derived, for the three materials
... Show MoreThe present study is concerned with Biostratigraphy of the Early-Middle Miocene outcrops of Jeribe Formation in the Zurbatiyah area, Wasit Governorate, Eastern Iraq. Forty-two Samples collected from Shur Sharin and AL-Hashima outcrop sections. The fossil content is rich in large and small benthic foraminifera; Twenty-one species and genus are identified in this study, in addition to coral, gastropoda, pelecypoda, ostracoda, alge, echinoid and shell fragments. According to the presence of benthic foraminifera, two Biozone have been identified in the Jeribe: Austrotrillina asmariensis-Dendritina rangi Concurrent Zone and Borelis melo curdica range zone.The age of the Formation determined as Early-Middle Miocene depending on these Bioz
... Show More