Memory Reduction Research Articles

Design and Performance Analysis of Hardware Realization of 3GPP Physical Layer for 5G Cell Search

5G Cell Search (CS) is the first step for user equipment (UE) to initiate communication with the 5G node B (gNB) every time it is powered ON. In cellular networks, CS is accomplished via synchronization signals (SS) broadcasted by gNB. 5G 3rd generation partnership project (3GPP) specifications offer a detailed discussion on the SS generation at gNB, but a limited understanding of their blind search and detection is available. Unlike 4G, 5G SS may not be transmitted at the center of carrier frequency, and their frequency location is unknown to UE. In this work, we demonstrate the 5G CS by designing 3GPP compatible hardware realization of the physical layer (PHY) of the gNB transmitter and UE receiver. The proposed SS detection explores a novel down-sampling approach resulting in a 60% reduction in on-chip memory and 50% lower search time. Via detailed performance analysis, we analyze the functional correctness, computational complexity, and latency of the proposed approach for different word lengths, signal-to-noise ratio (SNR), and down-sampling factors. We demonstrate end-to-end 5G CS using GNU Radio-based RFNoC framework on the USRP-FPGA platform and achieve 66% faster SS search compared to software. The 3GPP compatibility and demonstration on hardware strengthen the commercial significance of the proposed work.

OR01-05 Growth Hormone Action In Somatostatin Neurons Regulates Anxiety And Fear Memory

Abstract Disclosure: W.O. dos Santos: None. V.A. Juliano: None. H.R. Vieira: None. R. Frazao: None. C.D. Munhoz: None. J. Donato: None. Dysfunctions in growth hormone (GH) secretion increase the prevalence of anxiety and depression. Furthermore, GH receptor (GHR) signaling in the amygdala has been associated with fear memory, a key feature of post-traumatic stress disorder. However, it is currently unknown which neuronal population is targeted by GH action to influence neuropsychiatric diseases. Here we showed that approximately 60% of somatostatin (SST)-expressing neurons in the extended amygdala are directly responsive to GH. GHR ablation in SST-expressing cells (SSTΔGHR mice) led to increased anxiety in the light-dark box and elevated plus maze tests, whereas SSTΔGHR females showed no changes in anxiety, compared to control animals. Using auditory Pavlovian fear conditioning, both male and female SSTΔGHR mice exhibited a significant reduction in fear memory. In contrast, GHR ablation in SST neurons did not affect memory in the novel object recognition test. The gene expression was analyzed in a bilateral micropunch comprising the central nucleus of the amygdala and basolateral complex of male mice. Increased mRNA expression of Gabbr2 (GABAB2 receptor), Syp (synaptophysin), and Bdnf were observed in the amygdala of SSTΔGHR mice. In conclusion, GHR expression in SST neurons is necessary to regulate anxiety in males, but not in female mice. GHR ablation in SST neurons also decreases fear memory and affects the gene expression in the amygdala. Our findings identified for the first time a specific neuronal population responsible for mediating the effects of GH on behavioral aspects associated with neuropsychiatric diseases. Presentation: Thursday, June 15, 2023

SAT605 Growth Hormone Action In Somatostatin Neurons Regulates Anxiety And Fear Memory

Abstract Disclosure: W.O. dos Santos: None. V.A. Juliano: None. H.R. Vieira: None. R. Frazao: None. C.D. Munhoz: None. J. Donato: None. Dysfunctions in growth hormone (GH) secretion increase the prevalence of anxiety and depression. Furthermore, GH receptor (GHR) signaling in the amygdala has been associated with fear memory, a key feature of post-traumatic stress disorder. However, it is currently unknown which neuronal population is targeted by GH action to influence neuropsychiatric diseases. Here we showed that approximately 60% of somatostatin (SST)-expressing neurons in the extended amygdala are directly responsive to GH. GHR ablation in SST-expressing cells (SSTΔGHR mice) led to increased anxiety in the light-dark box and elevated plus maze tests, whereas SSTΔGHR females showed no changes in anxiety, compared to control animals. Using auditory Pavlovian fear conditioning, both male and female SSTΔGHR mice exhibited a significant reduction in fear memory. In contrast, GHR ablation in SST neurons did not affect memory in the novel object recognition test. The gene expression was analyzed in a bilateral micropunch comprising the central nucleus of the amygdala and basolateral complex of male mice. Increased mRNA expression of Gabbr2 (GABAB2 receptor), Syp (synaptophysin), and Bdnf were observed in the amygdala of SSTΔGHR mice. In conclusion, GHR expression in SST neurons is necessary to regulate anxiety in males, but not in female mice. GHR ablation in SST neurons also decreases fear memory and affects the gene expression in the amygdala. Our findings identified for the first time a specific neuronal population responsible for mediating the effects of GH on behavioral aspects associated with neuropsychiatric diseases. Presentation: Saturday, June 17, 2023

Efficient block contrastive learning via parameter-free meta-node approximation

Contrastive learning has recently achieved remarkable success in many domains including graphs. However contrastive loss, especially for graphs, requires a large number of negative samples which is unscalable and computationally prohibitive with a quadratic time complexity. Sub-sampling is not optimal. Incorrect negative sampling leads to sampling bias. In this work, we propose a meta-node based approximation technique that is (a) simple, (b) canproxy all negative combinations (c) in quadratic cluster size time complexity, (d) at graph level, not node level, and (e) exploit graph sparsity. By replacing node-pairs with additive cluster-pairs, we compute the negatives in cluster-time at graph level. The resulting Proxy approximated meta-node Contrastive (PamC) loss, based on simple optimized GPU operations, captures the full set of negatives, yet is efficient with a linear time complexity. By avoiding sampling, we effectively eliminate sample bias. We meet the criterion for larger number of samples, thus achieving block-contrastiveness, which is proven to outperform pair-wise losses. We use learnt soft cluster assignments for the meta-node construction, and avoid possible heterophily and noise added during edge creation. Theoretically, we show that real world graphs easily satisfy conditions necessary for our approximation. Empirically, we show promising accuracy gains over state-of-the-art graph clustering on 6 benchmarks. Importantly, we gain substantially in efficiency; over 2x reduction in training time and over 5x in GPU memory reduction. Additionally, our embeddings, combined with a single learnt linear transformation, is sufficient for node classification; we achieve state-of-the-art on Citeseer classification benchmark. code:https://github.com/gayanku/PAMC

Adaptive wave-particle decomposition in UGKWP method for high-speed flow simulations

With wave-particle decomposition, a unified gas-kinetic wave-particle (UGKWP) method has been developed for multiscale flow simulations. With the variation of the cell Knudsen number, the UGKWP method captures the transport process in all flow regimes without the kinetic solver’s constraint on the numerical mesh size and time step being determined by the kinetic particle mean free path and particle collision time. In the current UGKWP method, the cell Knudsen number, which is defined as the ratio of particle collision time to numerical time step, is used to distribute the components in the wave-particle decomposition. The adaptation of particles in the UGKWP method is mainly for the capturing of the non-equilibrium transport. In this aspect, the cell Knudsen number alone is not enough to identify the non-equilibrium state. For example, in the equilibrium flow regime with a Maxwellian distribution function, even at a large cell Knudsen number, the flow evolution can be still modelled by the Navier-Stokes solver. More specifically, in the near space environment both the hypersonic flow around a space vehicle and the plume flow from a satellite nozzle will encounter a far field rarefied equilibrium flow in a large computational domain. In the background dilute equilibrium region, the large particle collision time and a uniform small numerical time step can result in a large local cell Knudsen number and make the UGKWP method track a huge number of particles for the far field background flow in the original approach. But, in this region the analytical wave representation can be legitimately used in the UGKWP method to capture the nearly equilibrium flow evolution. Therefore, to further improve the efficiency of the UGKWP method for multiscale flow simulations, an adaptive UGKWP (AUGKWP) method is developed with the introduction of an additional local flow variable gradient-dependent Knudsen number. As a result, the wave-particle decomposition in the UGKWP method is determined by both the cell and gradient Knudsen numbers, and the use of particles in the UGKWP method is solely to capture the non-equilibrium flow transport. The current AUGKWP method becomes much more efficient than the previous one with the cell Knudsen number only in the determination of wave-particle composition. Many numerical tests, including Sod shock tube, normal shock structure, hypersonic flow around cylinder, flow around reentry capsule, and an unsteady nozzle plume flow, have been conducted to validate the accuracy and efficiency of the AUGKWP method. Compared with the original UGKWP method, the AUGKWP method achieves the same accuracy, but has advantages in memory reduction and computational efficiency in the simulation for flows with the co-existing of multiple regimes.

An Eulerian based geometry conforming grid-block dynamic mesh refinement for the lattice Boltzmann method

An algorithm for a three-dimensional fine-grid block encompassing a moving solid body using a multi-relaxation time model of the lattice Boltzmann method is proposed and developed. In this numerical framework, a geometry-conforming fine block shifts a coarse lattice unit as the object translates by same amount. The effective utilization of higher-order spatial interpolation has been demonstrated in place of the commonly used low-order temporal interpolation in traditional grid refinement techniques within the lattice Boltzmann method. The successful application of this method has been showcased through three distinct cases: the settling of a solid sphere in a fluid tank under the influence of gravity, hovering motion of an elliptic airfoil, and the “clap and fling” motion of an insect wing. In this regard, various interpolation schemes based on the location of nodes in the overlapping zones of fine and coarse block are discussed. In addition, two cases were evaluated, case 1 where distributions are interpolated and case 2 where macroscopic variables are directly interpolated. It was observed that both interpolations gave same computational accuracy for low Reynolds number [∼O(102)]. However, as Re is increased [∼O(103)], direct interpolation of macroscopic variables proved erroneous and resulted in a large deviation in fluid forces and is not recommended. The geometry-conforming dynamic mesh refinement results in a substantial decrease in computation time, approximately 90%, along with a memory reduction of about 80% compared to the fully refined counterpart.

AoCStream: All-on-Chip CNN Accelerator with Stream-Based Line-Buffer Architecture and Accelerator-Aware Pruning.

Convolutional neural networks (CNNs) play a crucial role in many EdgeAI and TinyML applications, but their implementation usually requires external memory, which degrades the feasibility of such resource-hungry environments. To solve this problem, this paper proposes memory-reduction methods at the algorithm and architecture level, implementing a reasonable-performance CNN with the on-chip memory of a practical device. At the algorithm level, accelerator-aware pruning is adopted to reduce the weight memory amount. For activation memory reduction, a stream-based line-buffer architecture is proposed. In the proposed architecture, each layer is implemented by a dedicated block, and the layer blocks operate in a pipelined way. Each block has a line buffer to store a few rows of input data instead of a frame buffer to store the whole feature map, reducing intermediate data-storage size. The experimental results show that the object-detection CNNs of MobileNetV1/V2 and an SSDLite variant, widely used in TinyML applications, can be implemented even on a low-end FPGA without external memory.

Model order reduction methods applied to neural network training

AbstractNeural networks have emerged as powerful and versatile tools in the field of deep learning. As the complexity of the task increases, so do size and architectural complexity of the network, causing compression techniques to become a focus of current research. Parameter truncation can provide a significant reduction in memory and computational complexity. Originating from a model order reduction framework, the Discrete Empirical Interpolation Method is applied to the gradient descent training of neural networks and analyze for important parameters. The approach for various state‐of‐the‐art neural networks is compared to established truncation methods. Further metrics like L2 and Cross‐Entropy Loss, as well as accuracy and compression rate are reported.

Identification of biomarkers in patients with rheumatoid arthritis responsive to DMARDs but with progressive bone erosion.

Rheumatoid arthritis (RA) is an inflammatory autoimmune disease that may cause joint destruction and disability. The pharmacological treatment of RA aims at obtaining disease remission by effectively ceasing joint inflammation and arresting progressive bone erosions. Some patients present bone lesions accrual even after controlling joint inflammation with current therapies. Our study aimed to analyze lymphocyte subsets and levels of circulating cytokines in patients with RA with progressive bone erosions. We enrolled 20 patients with a diagnosis of RA and 12 healthy donors (HD). Patients with RA were divided into patients with bone erosions (RA-BE+) and without bone erosions (RA-BE-). Lymphocyte subsets in peripheral blood were evaluated by flow cytometry. Circulating cytokines levels were evaluated by protein array. The distribution of lymphocyte subsets was not able to separate HD from AR patients and RA-BE+ and RA-BE- in cluster analysis. We observed a significant expansion of CXCR5- PD1+ T peripheral helper cells (Tph cells) and a reduction in both total memory B cells and switched memory B cells in RA patients compared to HD. We observed an expansion in the frequency of total B cells in RA-BE+ patients compared to RA-BE- patients. Unsupervised hierarchical clustering analysis of 39 cytokines resulted in a fairly good separation of HD from RA patients but not of RA-BE+ patients from RA-BE- patients. RA-BE+ patients showed significantly higher levels of IL-11 and IL-17A than RA-BE- patients. We show that patients with progressive erosive disease are characterized by abnormalities in B cells and in cytokines with a proven role in bone reabsorption. Understanding the role played by B cells and the cytokine IL-11 and IL-17A in progressive erosive disease can help identify novel biomarkers of erosive disease and design treatment approaches aimed at halting joint damage in RA.

Age-related reductions in arousal-enhanced memory are moderated by trait emotion regulation

Emotional arousal is known to enhance episodic memory in young adults. However, compared to valence, little is known about how healthy aging impacts arousal-enhanced memory effects. Furthermore, while emotion regulation is believed to improve with age, it is unclear how individual differences in emotion regulation influence arousal-enhanced memory. In this large-scale online study, we investigated the impact of age and individual differences in emotion regulation on arousal-enhanced memory. During encoding, participants made arousal ratings about negative, neutral, and positive images, and we compared their subsequent memory of high and low-arousal images. We found the impact of emotional arousal on memory was reduced with age, especially for older adults who habitually suppress their emotions. Our findings show that arousal-related memory benefits are reduced with advancing age, and that individual differences in habitual usage of emotion regulation impact these age-related alterations.

Distinct Circuits From the Central Lateral Amygdala to the Ventral Part of the Bed Nucleus of Stria Terminalis Regulate Different Fear Memory

BackgroundThe ability to differentiate stimuli that predict fear is critical for survival; however, the underlying molecular and circuit mechanisms remain poorly understood. MethodsWe combined transgenic mice, in vivo transsynaptic circuit-dissecting anatomical approaches, optogenetics, pharmacological methods, and electrophysiological recording to investigate the involvement of specific extended amygdala circuits in different fear memory. ResultsWe identified the projections from central lateral amygdala (CeL) protein kinase C δ (PKCδ)–positive neurons and somatostatin (SST)-positive neurons to GABAergic (gamma-aminobutyric acidergic) and glutamatergic neurons in the ventral part of the bed nucleus of stria terminalis (vBNST). Prolonged optogenetic activation or inhibition of the PKCδCeL-vBNST pathway specifically reduced context fear memory, whereas the SSTCeL-vBNST pathway mainly reduced tone fear memory. Intriguingly, optogenetic manipulation of vBNST neurons that received the projection from PKCδCeL neurons exerted bidirectional regulation of context fear, whereas manipulation of vBNST neurons that received the projection from SSTCeL neurons could bidirectionally regulate both context and tone fear memory. We subsequently demonstrated the presence of δ and κ opioid receptor protein expression within the CeL-vBNST circuits, potentially accounting for the discrepancy between prolonged activation of GABAergic circuits and inhibition of downstream vBNST neurons. Finally, administration of an opioid receptor antagonist cocktail on the PKCδCeL-vBNST or SSTCeL-vBNST pathway successfully restored context or tone fear memory reduction induced by prolonged activation of the circuits. ConclusionsTogether, these findings establish a functional role for distinct CeL-vBNST circuits in the differential regulation and appropriate maintenance of fear.

An FPGA Design with High Memory Efficiency and Decoding Performance for 5G LDPC Decoder

A hardware-efficient implementation of a Low-Density Parity-Check (LDPC) decoder is presented in this paper. The proposed decoder design is based on the Hybrid Offset Min-Sum (HOMS) algorithm. In the check node processing of this decoder, only the first minimum is computed instead of the first two minimum values among all the variable-to-check message inputs as in the conventional approach. Additionally, taking advantage of the unique structure of 5G LDPC codes, layered scheduling and partially parallel structures are employed to minimize hardware costs. Implementation results on the Xilinx Kintex UltraScale+ FPGA platform show that the proposed decoder can achieve a throughput of 2.82 Gbps for 10 decoding iterations with a 5G LDPC codelength of 8832 bits and a code rate of 1/2. Moreover, it yields a check node memory reduction of 10% with respect to the baseline and provides a hardware usage efficiency of 4.96 hardware resources/layer/Mbps, while providing a decoding performance of 0.15 dB better than some of the existing decoders.

Growth Hormone Action in Somatostatin Neurons Regulates Anxiety and Fear Memory.

Dysfunctions in growth hormone (GH) secretion increase the prevalence of anxiety and other neuropsychiatric diseases. GH receptor (GHR) signaling in the amygdala has been associated with fear memory, a key feature of posttraumatic stress disorder. However, it is currently unknown which neuronal population is targeted by GH action to influence the development of neuropsychiatric diseases. Here, we showed that approximately 60% of somatostatin (SST)-expressing neurons in the extended amygdala are directly responsive to GH. GHR ablation in SST-expressing cells (SSTΔGHR mice) caused no alterations in energy or glucose metabolism. Notably, SSTΔGHR male mice exhibited increased anxiety-like behavior in the light-dark box and elevated plus maze tests, whereas SSTΔGHR females showed no changes in anxiety. Using auditory Pavlovian fear conditioning, both male and female SSTΔGHR mice exhibited a significant reduction in fear memory. Conversely, GHR ablation in SST neurons did not affect memory in the novel object recognition test. Gene expression was analyzed in a micro punch comprising the central nucleus of the amygdala (CEA) and basolateral (BLA) complex. GHR ablation in SST neurons caused sex-dependent changes in the expression of factors involved in synaptic plasticity and function. In conclusion, GHR expression in SST neurons is necessary to regulate anxiety in males, but not female mice. GHR ablation in SST neurons also decreases fear memory and affects gene expression in the amygdala, although marked sex differences were observed. Our findings identified for the first time a neurochemically-defined neuronal population responsible for mediating the effects of GH on behavioral aspects associated with neuropsychiatric diseases.SIGNIFICANCE STATEMENT Hormone action in the brain regulates different neurological aspects, affecting the predisposition to neuropsychiatric disorders, like depression, anxiety, and posttraumatic stress disorder. Growth hormone (GH) receptor is widely expressed in the brain, but the exact function of neuronal GH action is not fully understood. Here, we showed that mice lacking the GH receptor in a group of neurons that express the neuropeptide somatostatin exhibit increased anxiety. However, this effect is only observed in male mice. In contrast, the absence of the GH receptor in somatostatin-expressing neurons decreases fear memory, a key feature of posttraumatic stress disorder, in males and females. Thus, our study identified a specific group of neurons in which GH acts to affect the predisposition to neuropsychiatric diseases.

Refinement of the Barnes and Morris water maze protocols improves characterization of spatial cognitive deficits in the lithium-pilocarpine rat model of epilepsy

Temporal lobe epilepsy (TLE) often causes cognitive impairment, especially a decline in spatial memory. Reductions in spatial memory and learning are also common in rodent models of TLE. The Morris water maze and the Barnes maze are the standard methods for evaluating spatial learning and memory in rodents. However, animals with TLE may exhibit agitation, distress, and fail to follow the paradigmatic context of these tests, making the interpretation of experimental data difficult. This study optimized the procedure of the Morris water maze and the Barnes maze to evaluate spatial learning and memory in rats with the lithium-pilocarpine TLE model (LPM rats). It was demonstrated that LPM rats required a mandatory and prolonged habituation stage for both tests. Therefore, the experimental rats performed relatively well on these tests. Nevertheless, LPM rats exhibited a slower learning process compared to the control rats. LPM rats also showed a reduction in spatial memory formation. This was more pronounced in the Barnes maze. Also, LPM rats utilized a sequential strategy for searching in the Barnes maze and were incapable of developing a more efficient spatial search strategy that is common in control animals. The Barnes maze may be a better choice for assessing search strategies, learning deficits, and spatial memory in rats with TLE when choosing between the two tests. This is because of the risk of unexpected seizure occurrence during the Morris water maze tests, and the potential risks for animal welfare.

Temperature-drift effect analysis of microstrip filters based on parallel high-order DGTD and FETD method with memory reduction technique

The temperature-drift effect of microstrip filters is analyzed by the proposed electromagnetic-thermal co-simulation method based on parallel high-order discontinuous Galerkin time domain (DGTD) and finite-element time-domain (FETD) algorithms with a memory reduction technique. The element matrices of DGTD method are factorized into the product of coefficient matrices and universal matrices. The coefficient matrices are different for each tetrahedral element and need to be stored for all the elements. While the universal matrices are the same for all the tetrahedral elements and only need to be stored for one element. Since the sizes of the coefficient matrices are much smaller than the element matrices of DGTD method, the proposed method avoids storing the large element matrices and greatly reduce the memory requirement of the DGTD method. Thus it reduces the memory requirement of the whole electromagnetic-thermal co-simulation which is dominated by the memory consumption of the DGTD method. Large-scale parallel technique is adopted to accelerate the process of electromagnetic-thermal co-simulation. The proposed method provides a very powerful tool for temperature-drift effect analysis of microstrip filters.

Exercise therapy to prevent and treat Alzheimer's disease.

Alzheimer's disease (AD) is a progressive neurodegenerative disease in the elderly with dementia, memory loss, and severe cognitive impairment that imposes high medical costs on individuals. The causes of AD include increased deposition of amyloid beta (Aβ) and phosphorylated tau, age, mitochondrial defects, increased neuroinflammation, decreased synaptic connections, and decreased nerve growth factors (NGF). While in animals moderate-intensity exercise restores hippocampal and amygdala memory through increased levels of p-AKT, p-TrkB, and p-PKC and decreased levels of Aβ, tau phosphorylation, and amyloid precursor proteins (APP) in AD. Aerobic exercise (with an intensity of 50-75% of VO2 max) prevents hippocampal volume reduction, spatial memory reduction, and learning reduction through increasing synaptic flexibility. Exercise training induces the binding of brain-derived neurotrophic factor (BDNF) to TrkB and the binding of NGF to TrkA to induce cell survival and neuronal plasticity. After aerobic training and high-intensity interval training, the increase of VEGF, angiopoietin 1 and 2, NO, tPA, and HCAR1 in cerebral vessels causes increased blood flow and angiogenesis in the cerebellum, motor cortex, striatum, and hippocampus. In the hippocampus, exercise training decreases mitochondrial fragmentation, DRP1, and FIS1, improving OPA1, MFN1, MFN2, and mitochondrial morphology. In humans, acute exercise as an anti-inflammatory condition causes an acute increase in IL-6 and an increase in anti-inflammatory factors such as IL-1RA and IL-10. Moderate-intensity exercise also inhibits inflammatory markers such as IFN-γ, IL-1β, IL-6, CRP, TNF-α, sTNFR1, COX-2, and NF-κB. Aerobic exercise significantly increases plasma levels of BDNF, nerve growth factor, synaptic plasticity, motor activity, spatial memory, and exploratory behavior in AD subjects. Irisin is a myokine released from skeletal muscle during exercise and protects the hippocampus by suppressing Aβ accumulation and promoting hippocampal proliferation through STAT3 signaling. Therefore, combined exercise training such as aerobic training, strength training, balance and coordination training, and cognitive and social activities seems to provide important benefits for people with AD.

Wave‐based inversion at scale on graphical processing units with randomized trace estimation

AbstractThanks to continued performance improvements in software and hardware, wave‐equation‐based imaging technologies, such as full‐waveform inversion and reverse‐time migration, are becoming more commonplace. However, widespread adaptation of these advanced imaging modalities has not yet materialized because current implementations are not able to reap the full benefits from accelerators, in particular those offered by memory‐scarce graphics processing units. Through the use of randomized trace estimation, we overcome the memory bottleneck of this type of hardware. At the cost of limited computational overhead and controllable incoherent errors in the gradient, the memory footprint of adjoint‐state methods is reduced drastically. Thanks to this relatively simple to implement memory reduction via an approximate imaging condition, we are able to benefit from graphics processing units without memory offloading. We demonstrate the performance of the proposed algorithm on acoustic two‐ and three‐dimensional full‐waveform inversion examples and on the formation of image gathers in transverse tilted isotropic media.

Distinct B cell profiles characterise healthy weight and obesity pre- and post-bariatric surgery

Background/ObjectivesObesity-associated metabolic dysfunction and inflammation can be ameliorated by bariatric surgery. While obesity is also linked to impaired B cell activation, differentiation, and persistence in response to infection and vaccination little is known about post-operative immune B cell compartment and to what extent dysregulation in B cell pathways can be reversed. To bridge this gap in knowledge, we carried out in-depth evaluation of B cell composition in individuals with obesity prior to and following bariatric surgery compared to lean controls.Subjects/MethodsWe recruited individuals with obesity (BMI at least 35 kg/m2) before bariatric surgery (n = 21) and followed them up 6 months post-operatively (n = 17). As controls we recruited age- and sex-matched lean (BMI < 25) individuals (n = 18). We carried out comprehensive immunophenotyping of peripheral blood B cells as well as interrogated their association with inflammatory and metabolic parameters.ResultsIn obesity the balance of antigen-inexperienced and memory B cells in the peripheral blood is altered, with an expansion of naïve and a reduction in total memory B cells. 6 months following bariatric surgery this balance is restored. However, post-operative patients are uniquely characterised by an increase in B cell subsets associated with chronic inflammation – CD11c+CXCR5-IgD-CD27- double negative 2 (DN2) B cells and CD27+CD38++ plasmablasts. Correlations between B cells subsets, inflammatory and metabolic parameters were distinct in lean people and individuals with obesity pre- and post-bariatric surgery.ConclusionsBariatric surgery patients display a unique B cell profile 6 months post-operatively; this bears minimal resemblance to that of pre-operative patients and only partially overlaps with that of lean controls. Post-operative differences in the B cell compartment compared to lean controls are detected despite global amelioration of inflammation and restoration of metabolic health. Collectively, this indicates that bariatric surgery creates a specific immunometabolic state with potential implications for health outcomes.

Large depth-of-field ultra-compact microscope by progressive optimization and deep learning

The optical microscope is customarily an instrument of substantial size and expense but limited performance. Here we report an integrated microscope that achieves optical performance beyond a commercial microscope with a 5×, NA 0.1 objective but only at 0.15 cm3 and 0.5 g, whose size is five orders of magnitude smaller than that of a conventional microscope. To achieve this, a progressive optimization pipeline is proposed which systematically optimizes both aspherical lenses and diffractive optical elements with over 30 times memory reduction compared to the end-to-end optimization. By designing a simulation-supervision deep neural network for spatially varying deconvolution during optical design, we accomplish over 10 times improvement in the depth-of-field compared to traditional microscopes with great generalization in a wide variety of samples. To show the unique advantages, the integrated microscope is equipped in a cell phone without any accessories for the application of portable diagnostics. We believe our method provides a new framework for the design of miniaturized high-performance imaging systems by integrating aspherical optics, computational optics, and deep learning.

Balancing Tradeoffs in Network Queue Management Problem via Forward–Backward Sweeping with Finite Checkpoints

Network queue management can be modelled as an optimal control problem and is aimed at controlling the dropping rate, in which the state and control variables are the instantaneous queue length and the dropping rate, respectively. One way to solve it is by using an indirect method, namely forward–backward sweeping based on the Pontryagin minimum principle to derive control the trajectory of the dropping rate. However, there exists some performance balance issues in the network queue, such as memory usage versus runtime of the algorithm, or dropping rate versus network queue length. Many researchers have exploited symmetry for constrained systems, controllers, and model predictive control problems to achieve an exponential memory reduction and simple, intuitive optimal controllers. In this article, we introduce the integration of the checkpointing method into forward–backward sweeping to address such balancing issues. Specifically, we exploit the revolve algorithm in checkpointing and choose a finite number of checkpoints to reduce the complexity. Both numerical and simulation results in a popular network simulator (ns-2) are provided through two experiments: varying bandwidth and offered load, which solidify our proposal in comparison to other deployed queue management algorithms.