Explore the vast range of titles available.

This paper presents a novel approach for dynamically tuning the exploration parameter in Multi-Armed Bandit (MAB) algorithms using Deep Q-Networks (DQN), focusing on enhancing performance in static and dynamic environments. Traditional MAB algorithms such as Upper Confidence Bound (UCB) and Thompson Sampling (TS) rely on fixed exploration parameters and assume stationary reward distributions, limiting their effectiveness in real-world applications where reward distributions can be dynamic. This paper proposes a learning-based method where a DQN agent observes the state of the MAB environment and selects an appropriate exploration parameter from a predefined set to address this problem. Experimental results show that the DQN-enhanced UCB algorithm consistently outperforms its traditional counterpart in both static and dynamic environments by achieving lower cumulative regret. In contrast, DQN-tuned TS moderately improves dynamic settings but exhibits instability in static environments. These findings highlight the potential of integrating neural network-based learning with classical decision-making strategies to enable adaptive exploration in non-stationary environments, offering valuable insights for recommender systems and other sequential decision-making tasks.

To circumvent the von Neumann bottleneck, substantial progress has been made towards in-memory computing with synaptic devices. However, compact nanodevices implementing non-linear activation functions are required for efficient full-hardware implementation of deep neural networks. Here, we present an energy-efficient and compact Mott activation neuron based on vanadium dioxide and its successful integration with a conductive bridge random access memory (CBRAM) crossbar array in hardware. The Mott activation neuron implements the rectified linear unit function in the analogue domain. The neuron devices consume substantially less energy and occupy two orders of magnitude smaller area than those of analogue complementary metal–oxide semiconductor implementations. The LeNet-5 network with Mott activation neurons achieves 98.38% accuracy on the MNIST dataset, close to the ideal software accuracy. We perform large-scale image edge detection using the Mott activation neurons integrated with a CBRAM crossbar array. Our findings provide a solution towards large-scale, highly parallel and energy-efficient in-memory computing systems for neural networks. Energy- and area-efficient vanadium-dioxide-based Mott activation neuron devices enable the implementation of activation functions in neural networks.

CMOS-RRAM integration holds great promise for low energy and high throughput neuromorphic computing. However, most RRAM technologies relying on filamentary switching suffer from variations and noise, leading to computational accuracy loss, increased energy consumption, and overhead by expensive program and verify schemes. We developed a filament-free, bulk switching RRAM technology to address these challenges. We systematically engineered a trilayer metal-oxide stack and investigated the switching characteristics of RRAM with varying thicknesses and oxygen vacancy distributions to achieve reliable bulk switching without any filament formation. We demonstrated bulk switching at megaohm regime with high current nonlinearity, up to 100 levels without compliance current. We developed a neuromorphic compute-in-memory platform and showcased edge computing by implementing a spiking neural network for an autonomous navigation/racing task. Our work addresses challenges posed by existing RRAM technologies and paves the way for neuromorphic computing at the edge under strict size, weight, and power constraints. Filamentary RRAM technologies suffer from variations and noise, leading to computational accuracy loss, and increased energy consumption. Park et al. created a trilayer metal-oxide bulk switching RRAM technology without filament formation and showed edge computing for an autonomous navigation task.

Resistive RAM crossbar arrays offer an attractive solution to minimize off-chip data transfer and parallelize on-chip computations for neural networks. Here, we report a hardware/software co-design approach based on low energy subquantum conductive bridging RAM (CBRAM®) devices and a network pruning technique to reduce network level energy consumption. First, we demonstrate low energy subquantum CBRAM devices exhibiting gradual switching characteristics important for implementing weight updates in hardware during unsupervised learning. Then we develop a network pruning algorithm that can be employed during training, different from previous network pruning approaches applied for inference only. Using a 512 kbit subquantum CBRAM array, we experimentally demonstrate high recognition accuracy on the MNIST dataset for digital implementation of unsupervised learning. Our hardware/software co-design approach can pave the way towards resistive memory based neuro-inspired systems that can autonomously learn and process information in power-limited settings. To realize the potential of resistive RAM crossbar arrays as platforms for neuromorphic computing, reduced network-level energy consumption must be achieved. Here, the authors use a hardware/software co-design approach to realize reduced energy consumption during network training for the network.

Conventional rubber-modified asphalt typically suffers from low rubber content and requires high construction temperatures. This study developed a warm-mix high-content crumb-rubber-modified asphalt (CRMA) with an increased rubber particle content of over 20%; moreover, the optimization of the warm-mixing agent was determined. Its rutting and cracking resistance performances were investigated using a dynamic shear rheometer (DSR) and a bending beam rheometer (BBR). Fourier Transform Infrared (FTIR) and Atomic Force Microscopy (AFM) were used to characterize the aging resistance and microstructural characteristics. The key findings revealed that the optimal dosage of the SDYK-type warm-mix additive (SDYK; a surfactant used to improve the high-temperature stability, low-temperature crack resistance, and anti-aging performance of asphalt) was 0.6% for high-rubber-content CRMA. The combination of warm-mix additives and rubber granules enhanced the aging resistance and elasticity of the asphalt while also contributing to an increase in chemical functional group indicators. The decrease in both the aliphatic chain index and branched alkane index of CRMA indicates that the warm-mix agent and the rubber additive enhanced the aging resistance of the asphalt. The warm-mix agent reduced the roughness of the asphalt, counteracting the roughness-enhancing effect of crumb rubber. This was attributed to the lubrication effect induced by the water film during the mixing process, which promotes a more uniform distribution of the rubber crumb network. This research established a theoretical and experimental basis for the application of high-rubber-content CRMA in large-temperature-difference regions.

Inspired from the computational efficiency of the biological brain, spiking neural networks (SNNs) emulate biological neural networks, neural codes, dynamics, and circuitry. SNNs show great potential for the implementation of unsupervised learning using in-memory computing. Here, we report an algorithmic optimization that improves energy efficiency of online learning with SNNs on emerging non-volatile memory (eNVM) devices. We develop a pruning method for SNNs by exploiting the output firing characteristics of neurons. Our pruning method can be applied during network training, which is different from previous approaches in the literature that employ pruning on already-trained networks. This approach prevents unnecessary updates of network parameters during training. This algorithmic optimization can complement the energy efficiency of eNVM technology, which offers a unique in-memory computing platform for the parallelization of neural network operations. Our SNN maintains ~90% classification accuracy on the MNIST dataset with up to ~75% pruning, significantly reducing the number of weight updates. The SNN and pruning scheme developed in this work can pave the way toward applications of eNVM based neuro-inspired systems for energy efficient online learning in low power applications.

Electrophysiology is a decades-old technique widely used for monitoring activity of individual neurons and local field potentials. Optogenetics has revolutionized neuroscience studies by offering selective and fast control of targeted neurons and neuron populations. The combination of these two techniques is crucial for causal investigation of neural circuits and understanding their functional connectivity. However, electrical artifacts generated by light stimulation interfere with neural recordings and hinder the development of compact closed-loop systems for precise control of neural activity. Here, we demonstrate that transparent graphene micro-electrodes fabricated on a clear polyethylene terephthalate film eliminate the light-induced artifact problem and allow development of a compact battery-powered closed-loop optogenetics system. We extensively investigate light-induced artifacts for graphene electrodes in comparison to metal control electrodes. We then design optical stimulation module using micro-LED chips coupled to optical fibers to deliver light to intended depth for optogenetic stimulation. For artifact-free integration of graphene micro-electrode recordings with optogenetic stimulation, we design and develop a compact closed-loop system and validate it for different frequencies of interest for neural recordings. This compact closed-loop optogenetics system can be used for various applications involving optogenetic stimulation and electrophysiological recordings.

Conversion of astrocytes into neurons in vivo offers an alternative therapeutic approach for neuronal loss after injury or disease. However, not only the efficiency of the conversion of astrocytes into functional neurons by single Neurog2, but also the conundrum that whether Neurog2-induced neuronal cells (Neurog2-iNs) are further functionally integrated into existing matured neural circuits remains unknown. Here, we adopted the AAV(2/8) delivery system to overexpress single factor Neurog2 into astrocytes and found that the majority of astrocytes were successfully converted into neuronal cells in multiple brain regions, including the midbrain and spinal cord. In the midbrain, Neurog2-induced neuronal cells (Neurog2-iNs) exhibit neuronal morphology, mature electrophysiological properties, glutamatergic identity (about 60%), and synapse-like configuration local circuits. In the spinal cord, astrocytes from both the intact and lesioned sources could be converted into functional neurons with ectopic expression of Neurog2 alone. Notably, further evidence from our study also proves that Neurog2-iNs in the intact spinal cord are capable of responding to diverse afferent inputs from dorsal root ganglion (DRG). Together, this study does not merely demonstrate the feasibility of Neurog2 for efficient in vivo reprogramming, it gives an indication for the Neurog2-iNs as a functional and potential factor in cell-replacement therapy.

Cervical cancer (CC) is one of the most prevalent cancers worldwide. Single nucleotide polymorphisms (SNPs) of the epidermal growth factor receptor (EGFR) and epidermal growth factor (EGF) genes are associated with cancers in diverse populations; However, the roles of these genes in CC are uncertain. Associations between these SNPs and CC risk, as well as the risk of pathological type and clinical stage, were analysed. On thebasis of our data, a cross-cancer meta-analysis was performed to assess the roles of nine SNPs of the EGFR and EGF genes in cancer susceptibility. Finally, SNP‒SNP interactions were analysed in the present study. Our data showed that the potential SNP‒SNP interaction between EGFR and EGF may be associated with CC development in Chinese Han individuals. The meta-analysis indicated that these SNPs in EGFR and EGF may be associated with cancer risk, particularly in Asians. Cross-cancer SNP‒SNP interaction analysis demonstrated that the 9-SNP model exhibited significant synergistic effects in predicting cancer risk.

Mesoporous silica/polyaniline (PANI) composite nanofibers with a full interpenetrating structure were synthesized via a controlled sol–gel process and in situ polymerization route from anodic alumina oxide (AAO) membranes. In the synthetic process, AAO membranes act as a template for the preparation of PANI/silica composite fibers. After removal of the AAO templates by hydrochloric acid (HCl) dissolution, the as-prepared mesoporous silica/PANI composite fibers possess a uniform fiber morphology and full interpenetrating structure, resulting in a continuous conductive PANI network with high porosity and large specific surface area. SEM images of mesoporous silica/PANI composite fibers shows that the diameter of the silica fibers has not been changed before and after PANI deposition, proving that PANI is well filled in the channel. The nitrogen (N 2 ) absorption/desorption experiments suggest that the average pore size and the surface area of silica/PANI composite fibers are 5.7 nm and 147 m 2 /g respectively. The profile of isotherms is similar to that of blank silica nanofibers with lower adsorption capacity and relative pressure, which demonstrates that aniline can polymerize in the pore channel. The results of four-probe method show that the conductivity of the mesoporous silica/PANI composite fibers is 2.6 × 10 –4  S/cm. This synthetic strategy is also suitable for fabricating various nanostructured composite nanofibers of PANI and other inorganic materials. Graphical abstract