Pith. sign in

REVIEW 3 major objections 5 minor 88 references

Synergistic Development of Perovskite Memristors and Algorithms for Robust Analog Computing

T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper argues that analog neural networks on imperfect perovskite memristors can be made dependable by jointly optimizing the fabrication recipe with Bayesian optimization and training the network with a BO-tuned multinomial noise…

desk verdict A solid empirical co-optimization story undermined by a robustness theorem that doesn't match the test-time noise model. read the letter →

arxiv 2412.02779 v2 pith:43EWNABT submitted 2024-12-03 cs.LG cs.AIcs.ET

Nanyang Ye , Qiao Sun , Yifei Wang , Liujia Yang , Jundong Zhou , Lei Wang , Guang-Zhong Yang , Xinbing Wang
show 5 more authors
Chenghu Zhou Wei Ren Leilei Gu Huaqiang Wu Qinying Gu
This is my paper · ORCID
classification cs.LGcs.AIcs.ET
keywords perovskitememristorsanalogcomputingBayesianoptimizationnoiseinjectiondeepneuralnetworkrobustnessmemristorcrossbarin-memorydevicenon-idealities
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Perovskite memristors could make deep learning much more energy-efficient by doing matrix multiplication in analog hardware, but their imperfect, noisy conductance states wreck neural-network accuracy. This paper claims the problem can be solved by optimizing the device and the training algorithm together: a Bayesian-optimization loop first finds fabrication conditions that maximize a measurable 'usability' score, and a second Bayesian loop tunes a multinomial noise-injection scheme, BayesMulti, that makes networks tolerate the residual hardware imperfections. The authors report that the combined approach keeps high accuracy across tasks ranging from MNIST and CIFAR-10 to autonomous-driving point-cloud detection, antibody and glycan classification, and MiniGPT-4, and that a fabricated 10×10 perovskite crossbar loses only about 15% accuracy in hardware inference versus roughly 45% for standard training. They also present a theorem giving a bound on how large a multiplicative weight disturbance can be while the network's prediction stays consistent.

What carries the argument

The load-bearing objects are the usability score and the BayesMulti noise distribution. Usability is a single scalar that converts a measured I-V curve into a device-quality target: the length of the longest conductance increasing subsequence relative to a required minimum, discounted by $e^{-\sigma}$ for stochastic cycle-to-cycle variation. BayesMulti is a training-time randomization in which each weight is multiplied by $\eta$ with density $p_1$ at $\eta=0$, $p_2$ at $\eta=0.5$, and $1-p_1-p_2$ at $\eta=1$; Bayesian optimization with a Gaussian-process surrogate and expected improvement tunes the noise parameters. The argument is carried by Theorem 1, a functional-optimization bound: minimizing the worst-case perturbed prediction over all bounded $[0,1]$-valued functions produces the robustness radius $r$ stated above. The mechanism is that noise injection randomizes over a neighborhood of the parameter point, and the 0.5 multiplier immunizes the network against the same perturbation when it is applied at inference time.

What would settle it

Run BayesMulti with a noise model where each weight is multiplied by a continuous random factor whose support lies inside the claimed robustness set, such as log-normal noise with $\sigma$ matched to device measurements; if any such perturbation pushes the randomized network's score at or below the decision threshold while the perturbation magnitude lies within the radius $r$ from Theorem 1, the claim that prediction outcomes remain consistent is false for the actual noise model. Alternatively, compute the exact worst-case perturbed prediction for a small two-layer network by solving the optimization over the set $B$, and check whether the minimizer really occurs at $m=0$, $l=\Theta-r$ as the proof assumes.

Watch

Extended reading notes

Core claim

The central claim is that analog neural networks can be made robust to perovskite memristor non-idealities by jointly optimizing fabrication and training noise. On the device side, the paper defines usability as $\mathrm{Usability} = (l_{\mathrm{LCIS}}/\mathrm{Require\_len})e^{-\sigma}$, where $l_{\mathrm{LCIS}}$ is the length of the longest monotonically increasing segment of the conductance curve, $\mathrm{Require\_len}$ is a minimum working range, and $\sigma$ is the cycle-to-cycle log-normal variability of conductance; Bayesian optimization over perovskite type, nanowire length and diameter, lead electrodeposition time, and Ag thickness raised usability from 0.36 to 0.93 in twelve iterations. On the algorithm side, BayesMulti multiplies each weight by an independent random factor $\eta$ drawn from $\{0, 0.5, 1\}$ with probabilities $p_1$ and $p_2$, with those probabilities chosen by Bayesian optimization. The paper's Theorem 1 states that, for a binary classifier whose network output is $f_{\pi_0}(\theta_0) = \mathbb{E}_{\eta\sim\pi_0}[f(\theta_0 * \eta)]$, the maximum allowable multiplicative disturbance $r$ satisfies $r \le [\ln(1.5 - f_{\pi_0}(\theta_0)) - \Theta \ln(1-p_2)]/[\ln p_1 - \ln(1-p_2)]$, where $\Theta$ is the number of parameters; if correct, any perturbation within that set leaves the prediction on the same side of the decision threshold. This is, to the authors' knowledge, the first demonstration of analog-computing inference on a large vision-language model, and the hardware test on a real perovskite crossbar supports the accuracy-stability claim.

Load-bearing premise

The theoretical robustness guarantee assumes the only perturbations that matter are coordinate-wise multiplications by exactly 0, 0.5, or 1, while the noisy hardware is modeled with continuous log-normal multiplicative drift, so the guarantee does not literally cover the noise used in the experiments.

Editorial extensions

If this is right

  • If Theorem 1 is correct, a BayesMulti-trained binary classifier's prediction cannot flip for any multiplicative disturbance inside the stated set, so analog implementations can be certified against a bounded class of device drift.
  • BayesMulti-trained analog networks keep usable accuracy at much lower usability values than standard empirical-risk-minimization training, in some KITTI detection settings achieving 10 to 100 times the ERM accuracy.
  • The BO fabrication loop, starting from one initial configuration, reached a near-optimal perovskite recipe in twelve iterations, including parameter choices outside human expertise such as smaller nanowire diameters.
  • A real 10×10 perovskite memristor crossbar implementing inference on a moon-shaped dataset loses about 15% accuracy with BayesMulti versus about 45% with ERM, and the array-only energy efficiency is claimed to exceed a Tesla V100 by over 270 times, excluding ADC and peripheral circuitry.
  • The same noise-injection recipe transfers to deeper and wider networks, including PointPillar, Mason's CNN, SweetNet, and MiniGPT-4, indicating the approach is architecture- and task-agnostic.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The theoretical radius $r$ depends on the parameter count $\Theta$ through the term $-\Theta\ln(1-p_2)$, so for very large models the guaranteed perturbation set may shrink; a layer-wise or block-wise noise schedule is a natural extension the paper does not explore.
  • Because the theorem's proof considers perturbations whose coordinates are exactly 0, 0.5, or 1, while the simulations use continuous log-normal multipliers, the experimental results should be read as evidence for the method's practical robustness rather than as a direct validation of the theorem's bound.
  • The usability metric, being derived from basic I-V measurements, could serve as a standardized figure of merit for comparing memristive technologies beyond perovskites, provided its correlation with end-task accuracy is validated across independent labs.
  • The reported 270× energy advantage counts only the crossbar array; including analog-to-digital conversion and peripheral circuits will change the ratio, and the same training recipe could be tested on full-system benchmarks.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper proposes a synergistic methodology for co-optimizing perovskite memristor fabrication and the robustness of analog deep neural networks. Fabrication conditions are selected through Bayesian optimization using a 'usability' metric derived from measured I-V characteristics, and the training method 'BayesMulti' injects multinomial noise whose parameters are tuned via Bayesian optimization. The central theoretical claim is Theorem 1, which asserts that training with the multinomial noise guarantees consistent predictions under multiplicative parameter perturbations within a robustness set B, with a maximum allowable radius r given by Eq. (2). The method is evaluated on image classification, autonomous driving, biological sequence modeling, a large vision-language model, and a 10x10 perovskite memristor crossbar, with reported improvements over empirical risk minimization and an energy-efficiency comparison against a GPU.

Significance. The empirical scope is broad and the hardware demonstration on a real crossbar is a valuable contribution. If the theoretical guarantee were valid for the noise model actually tested, the work would be an important advance in analog co-design. The fabrication Bayesian optimization pipeline and the crossbar validation are strengths, and the paper provides concrete prototypes in Supplementary Note 4. However, the central robustness theorem does not apply to the continuous log-normal noise used in all experiments, and the stated bound appears numerically vacuous for networks of realistic size. The theoretical contribution therefore needs substantial revision before the paper's central claims can be accepted.

major comments (3)
  1. [Supplementary Note 3 / Methods Eq. (4)] The robustness analysis in Supplementary Note 3 (Lemma 1 and the derivation of Eq. (16)) is carried out for perturbations δ whose coordinates take values in {0, 0.5, 1}, with the counts k, l, m fully determining the DF bound. In contrast, the paper's own stochastic non-ideality model (Methods, Eq. (4)) is θ' ← θe^λ with λ ∼ N(0, σ²), a continuous multiplicative perturbation. For a log-normal draw, δ_i ∉ {0, 0.5, 1} for every coordinate with probability one, so the robustness set B (interpreted in the proof as Θ−m−l≤r) is violated for any r<Θ. Consequently, Theorem 1 does not cover the perturbations generated by PerovskiteMemSim and used in all experimental evaluations, leaving the abstract's claim of a theoretical guarantee for memristor non-idealities unsupported for the tested perturbation class.
  2. [Theorem 1, Eq. (2)] The statement of Theorem 1 is not well-formed: the robustness set is written as B = {δ : δ − 10 + δ − 0.50 − Θ ≤ r}, which is not a meaningful condition; the proof later interprets it as Θ−m−l≤r. In addition, Eq. (2) has the form r ≤ [ln(1.5−fπ0(θ0)) − Θ ln(1−p2)] / [ln p1 − ln(1−p2)]. Since p1 ≤ 1−p2, the denominator is non-positive, and for typical values (e.g., p1=0.3, p2=0.3, Θ=100, fπ0(θ0)=0.8) the right-hand side is negative. As printed, the theorem therefore yields no positive robustness radius for networks of the size used in the experiments, making the quantitative guarantee vacuous.
  3. [Supplementary Note 3, robustness analysis after Eq. (23)] The lower bound obtained by setting m=0 and l=Θ−r has the form fπ0(θ0) − 1 + p1^r(1−p2)^{Θ−r}. Because (1−p2)<1, the second term decays exponentially in Θ. For any network with more than a handful of parameters, fπ0(θ0) − 1 + p1^r(1−p2)^{Θ−r} > 0.5 cannot hold even for r=0 unless fπ0(θ0) is extremely close to 1.5, which is impossible since fπ0(θ0) ≤ 1. Thus the proof method, even under its own discrete-perturbation assumptions, cannot certify positive robustness for realistic network sizes; the quantitative claim in Eq. (2) needs to be revisited.
minor comments (5)
  1. [Theorem 1] The expression for the robustness set B contains obvious typographical errors and should be rewritten in terms of the count variables k, l, m used in the proof.
  2. [Methods Eq. (5) / Supplementary Note 4] The usability metric depends on the free parameter Require_len, which is set to 35 in the code but not justified in the main text; the authors should state how this threshold is chosen and whether the conclusions are sensitive to it.
  3. [Figure 6(b) caption] The caption says 'different hardware non-idealities (σ values)' while the horizontal axis is labeled usability; please make the caption consistent with the axis definition.
  4. [Introduction and Reference [30]] The prior work in Reference [30] already proposed a Bayes-optimized noise injection approach for analog DNNs; the text should clarify what BayesMulti adds beyond [30] and why the multinomial distribution is essential to the new method.
  5. [Abstract and Results] The abstract's 'up to 100-fold improvements' should be accompanied by a precise definition of the ratio being reported (e.g., accuracy ratio, error ratio, or energy efficiency), as the metric is not clear without reading the figures.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the central robustness guarantee is a conditional theorem derived from the multinomial training noise, and the empirical validation is anchored to external benchmarks and a real memristor crossbar.

full rationale

The claimed theoretical guarantee (Theorem 1, Eq. 2) is derived in Supplementary Note 3 from the injected multinomial noise distribution via a Lagrangian relaxation, and the bound is an explicit function of the noise parameters p1, p2, the parameter dimension Θ, and the noise-averaged model output fπ0(θ0). This is a conditional mathematical statement about the training procedure, not a fitted parameter renamed as a prediction, so it does not reduce to its inputs by construction. The evaluation pipeline is also not circular: PerovskiteMemSim maps weights using LCIS and log-normal drift estimated from measured I-V curves, and the paper includes external benchmarks (MNIST, CIFAR-10, KITTI, HIV, CoV AbDab, glycan data, LVLM tasks) plus a physical 10x10 perovskite crossbar demonstration. The robustness theorem is stated for a discrete ternary perturbation set δ ∈ {0, 0.5, 1}, whereas the paper's own device noise model (Methods Eq. 4) is continuous log-normal; that is a correctness and coverage gap, not a circular reduction. The only self-citation, [30], is used as background support for the importance of inductive noise and is not load-bearing for the present theorem or experiments. Accordingly, no circular step is identified, and the score reflects only a minor, non-load-bearing self-citation.

Assumptions & free parameters 3 free parameters · 4 assumptions · 1 invented entities

The central claims rest on a small set of hand-chosen thresholds and priors. The noise probabilities are tuned but not disclosed, the LCIS length threshold is arbitrary, and the usability formula is an ad hoc product. The log-normal noise assumption and independence across weights are borrowed from the ReRAM literature without fresh validation on perovskite devices.

free parameters (3)
  • p1, p2 (multinomial noise probabilities) = not reported (chosen by BO)
    Noise injection distribution in BayesMulti; values are tuned per task via Bayesian optimization but final values are not given in the text.
  • Require_len (minimum operative conductance length) = 35 (in code listing)
    Threshold for non-monotonic noise; chosen by hand, affects usability and simulated weight mapping.
  • Usability metric form = lLCIS/Require_len * exp(-sigma)
    The product of two heuristics is chosen as the optimization objective without a derivation from downstream task performance; used as both BO target and evaluation parameter.
assumptions (4)
  • domain assumption Weight drift is multiplicative log-normal (theta' = theta * e^lambda, lambda ~ N(0, sigma^2))
    Used in Eq. 4 to simulate stochastic non-ideality; cited from prior ReRAM literature but not validated for perovskite devices here.
  • domain assumption Parameter perturbations are i.i.d. across weights
    The robustness proof and simulation treat each weight independently; real memristor variations may be correlated across devices.
  • standard math Functional relaxation and min-max exchange in Theorem 2
    Used to derive the lower bound; assumes the relaxed function space and Lagrange dual are tight enough, which is not fully justified.
  • standard math Gaussian process surrogate and EI acquisition for BO
    Standard BO ingredients; assumes smoothness of the usability function over fabrication parameters.
invented entities (1)
  • Usability metric
    purpose: Combines monotonic and stochastic non-idealities into a single scalar for fabrication optimization and evaluation.
    A new composite indicator with no external benchmark; its predictive validity for downstream DNN accuracy is only demonstrated in the paper's own simulations.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Synergistic Development of Perovskite Memristors and Algorithms for Robust Analog Computing." pith.science (2026). https://pith.science/paper/43EWNABT

@misc{pith2026241202779,
  author       = {Pith},
  title        = {Pith review of: Synergistic Development of Perovskite Memristors and Algorithms for Robust Analog Computing},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/43EWNABT}},
  note         = {Machine review of arXiv:2412.02779}
}
abstract

Analog computing using non-volatile memristors has emerged as a promising solution for energy-efficient deep learning. New materials, like perovskites-based memristors are recently attractive due to their cost-effectiveness, energy efficiency and flexibility. Yet, challenges in material diversity and immature fabrications require extensive experimentation for device development. Moreover, significant non-idealities in these memristors often impede them for computing. Here, we propose a synergistic methodology to concurrently optimize perovskite memristor fabrication and develop robust analog DNNs that effectively address the inherent non-idealities of these memristors. Employing Bayesian optimization (BO) with a focus on usability, we efficiently identify optimal materials and fabrication conditions for perovskite memristors. Meanwhile, we developed "BayesMulti", a DNN training strategy utilizing BO-guided noise injection to improve the resistance of analog DNNs to memristor imperfections. Our approach theoretically ensures that within a certain range of parameter perturbations due to memristor non-idealities, the prediction outcomes remain consistent. Our integrated approach enables use of analog computing in much deeper and wider networks, which significantly outperforms existing methods in diverse tasks like image classification, autonomous driving, species identification, and large vision-language models, achieving up to 100-fold improvements. We further validate our methodology on a 10$\times$10 optimized perovskite memristor crossbar, demonstrating high accuracy in a classification task and low energy consumption. This study offers a versatile solution for efficient optimization of various analog computing systems, encompassing both devices and algorithms.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

88 extracted references · 68 canonical work pages

  1. [1]

    The process began with cutting 0.25 mm thick Aluminum (Al) foils into 20mm × 30 mm chips

    Device Fabrication PAM template fabrication To create Porous Anodic Alumina (PAM) for perovskite nanowire growth, we employed an anodic anodization method as previously described [45, 53, 74, 75]. The process began with cutting 0.25 mm thick Aluminum (Al) foils into 20mm × 30 mm chips. These chips were then flattened and sequentially cleaned with acetone ...

  2. [2]

    Electrical measurements The cyclic I-V characteristics were measured by Keithley 6487 with home-built LABVIEW programs

    Device Characterization SEM imaging The cross-sectional images of the PAM and perovskite NWs samples were collected by using a field emission scanning electron microscope ZEM15- Desktop in back-scattered electron (BSE) mode. Electrical measurements The cyclic I-V characteristics were measured by Keithley 6487 with home-built LABVIEW programs

  3. [3]

    usability

    Simulations of device non-idealities Simulating Non-monotonic Non-ideality In analog computing, neural networks weight ( θ) are represented by memristors’ conductance (C) which is modulated by the input trains of pulses. The non-idealities between conductance C and the number of charging pulses include non-linearity and non-monotonicity. Non-linearity can...

  4. [4]

    Pillar Feature Net: This is the first layer of PointPillar and is responsible for con- verting 3D point cloud data into a 2D representation of the column feature. It first divides the 3D space into a set of fixed-size columns, then calculates the features of the point cloud data in each column (such as maximum, minimum, average, etc.), and generates a fea...

  5. [5]

    Since the column features are already organized into 2D, standard 2D convolution operations can be applied directly

    2D Convolution Layers: After converting 3D point cloud data into 2D column features, PointPillar uses a series of 2D convolutional layers to process these fea- tures and extract the high-level features from them. Since the column features are already organized into 2D, standard 2D convolution operations can be applied directly. This greatly simplifies the...

  6. [6]

    This regression network can predict the objects’ position, size, and orientation in each bar

    Dense Head for 3D Object Detection: Finally, PointPillar uses a regression net- work to predict the bounding box of a 3D object. This regression network can predict the objects’ position, size, and orientation in each bar. In this way, the model can generate a 3D bounding box that can be used to locate and identify objects in the environment. In our exper...

  7. [7]

    Driverless car: Autonomous driving using deep reinforcement learning in urban environ- ment

    Abdur R Fayjie, Sabir Hossain, Doukhi Oualid, and Deok-Jin Lee. Driverless car: Autonomous driving using deep reinforcement learning in urban environ- ment. In 2018 15th international conference on ubiquitous robots (ur) , pages 896–901. IEEE, 2018

    2018

  8. [8]

    Policy optimization with demon- strations

    Bingyi Kang, Zequn Jie, and Jiashi Feng. Policy optimization with demon- strations. In International conference on machine learning , pages 2469–2478. PMLR, 2018

    2018

Show all 88 references
  1. [9]

    Learning to prompt for vision-language models

    Kaiyang Zhou, Jingkang Yang, Chen Change Loy, and Ziwei Liu. Learning to prompt for vision-language models. International Journal of Computer Vision, 130(9):2337–2348, 2022

    2022

  2. [10]

    Vinvl: Revisiting visual representations in vision-language models

    Pengchuan Zhang, Xiujun Li, Xiaowei Hu, Jianwei Yang, Lei Zhang, Lijuan Wang, Yejin Choi, and Jianfeng Gao. Vinvl: Revisiting visual representations in vision-language models. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 5579–5588, 2021

    2021

  3. [11]

    1.1 computing’s energy problem (and what we can do about it)

    Mark Horowitz. 1.1 computing’s energy problem (and what we can do about it). In 2014 IEEE international solid-state circuits conference digest of technical papers (ISSCC), pages 10–14. IEEE, 2014

    2014

  4. [12]

    Memory leads the way to better computing

    H-S Philip Wong and Sayeef Salahuddin. Memory leads the way to better computing. Nature Nanotechnology, 10(3):191–194, 2015

    2015

  5. [13]

    Memristive devices for computing

    J Joshua Yang, Dmitri B Strukov, and Duncan R Stewart. Memristive devices for computing. Nature nanotechnology, 8(1):13–24, 2013

    2013

  6. [14]

    The missing memristor found

    Dmitri B Strukov, Gregory S Snider, Duncan R Stewart, and R Stanley Williams. The missing memristor found. Nature, 453(7191):80–83, 2008

    2008

  7. [15]

    Memristors with diffusive dynamics as synaptic emulators for neuromorphic computing

    Zhongrui Wang, Saumil Joshi, Sergey E Savel’ev, Hao Jiang, Rivu Midya, Peng Lin, Miao Hu, Ning Ge, John Paul Strachan, Zhiyong Li, et al. Memristors with diffusive dynamics as synaptic emulators for neuromorphic computing. Nature Materials, 16(1):101–108, 2017

    2017

  8. [16]

    Stochastic phase-change neurons

    Tomas Tuma, Angeliki Pantazi, Manuel Le Gallo, Abu Sebastian, and Evan- gelos Eleftheriou. Stochastic phase-change neurons. Nature Nanotechnology, 11(8):693–699, 2016

    2016

  9. [17]

    A non-volatile organic electrochemical device as a low-voltage artificial synapse for neuromorphic computing

    Yoeri Van De Burgt, Ewout Lubberman, Elliot J Fuller, Scott T Keene, Gr´egorio C Faria, Sapan Agarwal, Matthew J Marinella, A Alec Talin, and Alberto Salleo. A non-volatile organic electrochemical device as a low-voltage artificial synapse for neuromorphic computing. Nature Ma...

    2017

  10. [18]

    Emerging nanoelectronic devices

    An Chen, James Hutchby, Victor Zhirnov, and George Bourianoff. Emerging nanoelectronic devices. John Wiley & Sons, USA, 2014

    2014

  11. [19]

    Brain-inspired computing via memory device physics

    D Ielmini, Z Wang, and Y Liu. Brain-inspired computing via memory device physics. APL Materials, 9(5), 2021

    2021

  12. [20]

    Nonideality-aware training for accurate and robust low-power memristive neural networks

    Dovydas Joksas, Erwei Wang, Nikolaos Barmpatsalos, Wing H Ng, Anthony J Kenyon, George A Constantinides, and Adnan Mehonic. Nonideality-aware training for accurate and robust low-power memristive neural networks. Advanced Science, 9(17):2105784, 2022. 48

    2022

  13. [21]

    Understanding memristive switching via in situ characterization and device modeling

    Wen Sun, Bin Gao, Miaofang Chi, Qiangfei Xia, J Joshua Yang, He Qian, and Huaqiang Wu. Understanding memristive switching via in situ characterization and device modeling. Nature Communications, 10(1):3453, 2019

    2019

  14. [22]

    Resistive switching materials for information processing

    Zhongrui Wang, Huaqiang Wu, Geoffrey W Burr, Cheol Seong Hwang, Kang L Wang, Qiangfei Xia, and J Joshua Yang. Resistive switching materials for information processing. Nature Reviews Materials, 5(3):173–195, 2020

    2020

  15. [23]

    Direct observa- tion of oxygen vacancy-driven structural and resistive phase transitions in la2/3sr1/3mno3

    Lide Yao, Sampo Inkinen, and Sebastiaan Van Dijken. Direct observa- tion of oxygen vacancy-driven structural and resistive phase transitions in la2/3sr1/3mno3. Nature Communications, 8(1):14544, 2017

    2017

  16. [24]

    Dot-product engine as computing memory to accelerate machine learning algorithms

    Miao Hu, John Paul Strachan, Zhiyong Li, R Stanley, et al. Dot-product engine as computing memory to accelerate machine learning algorithms. In 2016 17th International Symposium on Quality Electronic Design (ISQED) , pages 374–

    2016

  17. [25]

    In situ observation of filamentary conducting channels in an asymmetric ta2o5- x/tao2- x bilayer structure

    Gyeong-Su Park, Young Bae Kim, Seong Yong Park, Xiang Shu Li, Sung Heo, Myoung-Jae Lee, Man Chang, Ji Hwan Kwon, M Kim, U-In Chung, et al. In situ observation of filamentary conducting channels in an asymmetric ta2o5- x/tao2- x bilayer structure. Nature Communications, 4(1):2382, 2013

    2013

  18. [26]

    Observation of conducting filament growth in nanoscale resistive memories

    Yuchao Yang, Peng Gao, Siddharth Gaba, Ting Chang, Xiaoqing Pan, and Wei Lu. Observation of conducting filament growth in nanoscale resistive memories. Nature Communications, 3(1):732, 2012

    2012

  19. [27]

    Resistive switching of silicon-rich-oxide featur- ing high compatibility with cmos technology for 3d stackable and embedded applications

    Ru Huang, Lijie Zhang, Dejin Gao, Yue Pan, Shiqiang Qin, Poren Tang, Yimao Cai, and Yangyuan Wang. Resistive switching of silicon-rich-oxide featur- ing high compatibility with cmos technology for 3d stackable and embedded applications. Applied Physics A, 102:927–931, 2011

    2011

  20. [28]

    Silicon oxide: a non- innocent surface for molecular electronics and nanoelectronics studies

    Jun Yao, Lin Zhong, Douglas Natelson, and James M Tour. Silicon oxide: a non- innocent surface for molecular electronics and nanoelectronics studies. Journal of the American Chemical Society, 133(4):941–948, 2011

    2011

  21. [29]

    Mass transport in chalcogenide elec- trolyte films–materials and applications

    Michael N Kozicki and Maria Mitkova. Mass transport in chalcogenide elec- trolyte films–materials and applications. Journal of Non-crystalline Solids , 352(6-7):567–577, 2006

    2006

  22. [30]

    Resistive switching mechanism in ZnxCd1−xS nonvolatile memory devices

    Zheng Wang, Peter B Griffin, Jim McVittie, Simon Wong, Paul C McIntyre, and Yoshio Nishi. Resistive switching mechanism in ZnxCd1−xS nonvolatile memory devices. IEEE Electron Device Letters, 28(1):14–16, 2006

    2006

  23. [31]

    Recent advances in halide perovskite memristors: materials, structures, mechanisms, and applications

    Xinyu Xiao, Jing Hu, Sheng Tang, Kai Yan, Bo Gao, Hunglin Chen, and Dechun Zou. Recent advances in halide perovskite memristors: materials, structures, mechanisms, and applications. Advanced Materials Technologies, 5(6):1900914, 2020

    2020

  24. [32]

    Resistive switching behavior in organic–inorganic hybrid ch3nh3pbi3- xclx perovskite for resistive random access memory devices

    Eun Ji Yoo, Miaoqiang Lyu, Jung-Ho Yun, Chi Jung Kang, Young Jin Choi, and Lianzhou Wang. Resistive switching behavior in organic–inorganic hybrid ch3nh3pbi3- xclx perovskite for resistive random access memory devices. Advanced Materials, 27(40):6170–6175, 2015

    2015

  25. [33]

    A compute-in-memory chip based on resistive random-access memory

    Weier Wan, Rajkumar Kubendran, Clemens Schaefer, Sukru Burc Eryilmaz, Wenqiang Zhang, Dabin Wu, Stephen Deiss, Priyanka Raina, He Qian, Bin Gao, et al. A compute-in-memory chip based on resistive random-access memory. Nature, 608(7923):504–512, 2022. 49

    2022

  26. [34]

    A fully hardware-based memristive multilayer neural network

    Fatemeh Kiani, Jun Yin, Zhongrui Wang, J Joshua Yang, and Qiangfei Xia. A fully hardware-based memristive multilayer neural network. Science Advances, 7(48):eabj4801, 2021

    2021

  27. [35]

    Analog architectures for neural network acceleration based on non-volatile memory

    T Patrick Xiao, Christopher H Bennett, Ben Feinberg, Sapan Agarwal, and Matthew J Marinella. Analog architectures for neural network acceleration based on non-volatile memory. Applied Physics Reviews, 7(3), 2020

    2020

  28. [36]

    Improving the robustness of analog deep neural networks through a bayes-optimized noise injection approach

    Nanyang Ye, Linfeng Cao, Liujia Yang, Ziqing Zhang, Zhicheng Fang, Qinying Gu, and Guang-Zhong Yang. Improving the robustness of analog deep neural networks through a bayes-optimized noise injection approach. Communications Engineering, 2(1):25, 2023

    2023

  29. [37]

    Equivalent-accuracy accelerated neural-network training using analogue memory

    Stefano Ambrogio, Pritish Narayanan, Hsinyu Tsai, Robert M Shelby, Irem Boybat, Carmelo Di Nolfo, Severin Sidler, Massimo Giordano, Martina Bod- ini, Nathan CP Farinha, et al. Equivalent-accuracy accelerated neural-network training using analogue memory. Nature, 558(7708):60–67, 2018

    2018

  30. [38]

    A ti/alo x/tao x/pt analog synapse for memristive neural network

    Yi Sun, Hui Xu, Chao Wang, Bing Song, Haijun Liu, Qi Liu, Sen Liu, and Qingjiang Li. A ti/alo x/tao x/pt analog synapse for memristive neural network. IEEE Electron Device Letters, 39(9):1298–1301, 2018

    2018

  31. [39]

    Rescuing memristor-based neuromor- phic design with high defects

    C Liu, M Hu, JP Strachan, and H Li. Rescuing memristor-based neuromor- phic design with high defects. in2017 54th acm/edac/ieee design automation conference (dac) 2017 jun 18 (pp. 1-6)

    2017

  32. [40]

    In situ learning using intrinsic mem- ristor variability via markov chain monte carlo sampling

    Thomas Dalgaty, Niccolo Castellani, Cl ´ement Turck, Kamel-Eddine Harabi, Damien Querlioz, and Elisa Vianello. In situ learning using intrinsic mem- ristor variability via markov chain monte carlo sampling. Nature Electronics, 4(2):151–161, 2021

    2021

  33. [42]

    Multibit memory operation of metal- oxide bi-layer memristors

    Spyros Stathopoulos, Ali Khiat, Maria Trapatseli, Simone Cortese, Alexantrou Serb, Ilia Valov, and Themis Prodromakis. Multibit memory operation of metal- oxide bi-layer memristors. Scientific Reports, 7(1):17532, 2017

    2017

  34. [43]

    A review of resistive switching devices: performance improvement, characterization, and applications

    Tuo Shi, Rui Wang, Zuheng Wu, Yize Sun, Junjie An, and Qi Liu. A review of resistive switching devices: performance improvement, characterization, and applications. Small Structures, 2(4):2000109, 2021

    2021

  35. [44]

    Low-dimensional organic–inorganic halide perovskite: structure, properties, and applications

    Ravi K Misra, Bat-El Cohen, Lior Iagher, and Lioz Etgar. Low-dimensional organic–inorganic halide perovskite: structure, properties, and applications. ChemSusChem, 10(19):3712–3721, 2017

    2017

  36. [45]

    Organic and perovskite memristors for neuromorphic computing

    Hea-Lim Park and Tae-Woo Lee. Organic and perovskite memristors for neuromorphic computing. Organic Electronics, 98:106301, 2021

    2021

  37. [46]

    Compliance-free multileveled resistive switching in a transparent 2d perovskite for neuromorphic computing

    Mohit Kumar, Hong-Sik Kim, Dae Young Park, Mun Seok Jeong, and Joon- dong Kim. Compliance-free multileveled resistive switching in a transparent 2d perovskite for neuromorphic computing. ACS Applied Materials & Interfaces , 10(15):12768–12772, 2018

    2018

  38. [47]

    Ionotronic halide perovskite drift-diffusive synapses for low-power neuromorphic computation

    Rohit Abraham John, Natalia Yantara, Yan Fong Ng, Govind Narasimman, Edoardo Mosconi, Daniele Meggiolaro, Mohit R Kulkarni, Pradeep Kumar 50 Gopalakrishnan, Chien A Nguyen, Filippo De Angelis, et al. Ionotronic halide perovskite drift-diffusive synapses for low-power neuromorp...

    2018

  39. [48]

    Energy-efficient hybrid perovskite mem- ristors and synaptic devices

    Zhengguo Xiao and Jinsong Huang. Energy-efficient hybrid perovskite mem- ristors and synaptic devices. Advanced Electronic Materials , 2(7):1600100, 2016

    2016

  40. [49]

    Leaky integrate-and- fire neurons based on perovskite memristor for spiking neural networks

    Jia-Qin Yang, Ruopeng Wang, Zhan-Peng Wang, Qin-Yuan Ma, Jing-Yu Mao, Yi Ren, Xiaoyang Yang, Ye Zhou, and Su-Ting Han. Leaky integrate-and- fire neurons based on perovskite memristor for spiking neural networks. Nano Energy, 74:104828, 2020

    2020

  41. [50]

    Lead-free perovskite nanowire array photodetectors with drastically improved stability in nanoengineering templates

    Aashir Waleed, Mohammad Mahdi Tavakoli, Leilei Gu, Ziyi Wang, Daquan Zhang, Arumugam Manikandan, Qianpeng Zhang, Rongjun Zhang, Yu-Lun Chueh, and Zhiyong Fan. Lead-free perovskite nanowire array photodetectors with drastically improved stability in nanoengineering templates. N...

    2017

  42. [51]

    3d arrays of 1024-pixel image sensors based on lead halide perovskite nanowires

    Leilei Gu, Mohammad Mahdi Tavakoli, Daquan Zhang, Qianpeng Zhang, Aashir Waleed, Yiqun Xiao, Kwong-Hoi Tsui, Yuanjing Lin, Lei Liao, Jian- nong Wang, et al. 3d arrays of 1024-pixel image sensors based on lead halide perovskite nanowires. Advanced Materials, 28(44):9713–9721, 2016

    2016

  43. [52]

    Three-dimensional perovskite nanowire array–based ultrafast resistive ram with ultralong data retention

    Yuting Zhang, Swapnadeep Poddar, He Huang, Leilei Gu, Qianpeng Zhang, Yu Zhou, Shuai Yan, Sifan Zhang, Zhitang Song, Baoling Huang, et al. Three-dimensional perovskite nanowire array–based ultrafast resistive ram with ultralong data retention. Science Advances, 7(36):eabg3788, 2021

    2021

  44. [53]

    Benchmarking the performance of bayesian optimization across multiple experimental materials science domains.npj Computational Materials, 7(1):188, 2021

    Qiaohao Liang, Aldair E Gongora, Zekun Ren, Armi Tiihonen, Zhe Liu, Shi- jing Sun, James R Deneault, Daniil Bash, Flore Mekki-Berrada, Saif A Khan, et al. Benchmarking the performance of bayesian optimization across multiple experimental materials science domains.npj Computati...

    2021

  45. [54]

    Constrained bayesian optimization for automatic chemical design using variational autoencoders

    Ryan-Rhys Griffiths and Jos ´e Miguel Hern´andez-Lobato. Constrained bayesian optimization for automatic chemical design using variational autoencoders. Chemical Science, 11(2):577–586, 2020

    2020

  46. [55]

    Bayesian reaction optimization as a tool for chemical synthesis

    Benjamin J Shields, Jason Stevens, Jun Li, Marvin Parasram, Farhan Damani, Jesus I Martinez Alvarado, Jacob M Janey, Ryan P Adams, and Abigail G Doyle. Bayesian reaction optimization as a tool for chemical synthesis. Nature, 590(7844):89–96, 2021

    2021

  47. [56]

    Taking the human out of the loop: A review of bayesian optimization

    Bobak Shahriari, Kevin Swersky, Ziyu Wang, Ryan P Adams, and Nando De Freitas. Taking the human out of the loop: A review of bayesian optimization. Proceedings of the IEEE, 104(1):148–175, 2015

    2015

  48. [57]

    Pfra- zier2018tutorialractical bayesian optimization of machine learning algorithms

    Jasper Snoek, Hugo Larochelle, and Ryan P Adams. Pfra- zier2018tutorialractical bayesian optimization of machine learning algorithms. Advances in Neural Information Processing Systems, 25, 2012

    2012

  49. [58]

    A tutorial on bayesian optimization

    Peter I Frazier. A tutorial on bayesian optimization. arXiv preprint arXiv:1807.02811, 2018

    2018 arXiv

  50. [59]

    Down-scalable and ultra-fast memristors with ultra-high density three- dimensional arrays of perovskite quantum wires

    Swapnadeep Poddar, Yuting Zhang, Leilei Gu, Daquan Zhang, Qianpeng Zhang, Shuai Yan, Matthew Kam, Sifan Zhang, Zhitang Song, Weida Hu, 51 et al. Down-scalable and ultra-fast memristors with ultra-high density three- dimensional arrays of perovskite quantum wires. Nano Letters,...

    2021

  51. [60]

    The mnist database of handwritten digit images for machine learning research

    Li Deng. The mnist database of handwritten digit images for machine learning research. IEEE Signal Processing Magazine, 29(6):141–142, 2012

    2012

  52. [61]

    Cifar-10 (canadian institute for advanced research)

    Alex Krizhevsky, Vinod Nair, and Geoffrey Hinton. Cifar-10 (canadian institute for advanced research). URL http://www. cs. toronto. edu/kriz/cifar. html, 5(4):1, 2010

    2010

  53. [62]

    Vision meets robotics: The kitti dataset.The International Journal of Robotics Research, 32(11):1231–1237, 2013

    Andreas Geiger, Philip Lenz, Christoph Stiller, and Raquel Urtasun. Vision meets robotics: The kitti dataset.The International Journal of Robotics Research, 32(11):1231–1237, 2013

    2013

  54. [63]

    Hiv sequence compendium 2018

    Brian Thomas Foley, Bette Tina Marie Korber, Thomas Kenneth Leitner, Cris- tian Apetrei, Beatrice Hahn, Ilene Mizrachi, James Mullins, Andrew Rambaut, and Steven Wolinsky. Hiv sequence compendium 2018. 6 2018

    2018

  55. [64]

    Cov-abdab: the coronavirus antibody database

    Matthew IJ Raybould, Aleksandr Kovaltsuk, Claire Marks, and Charlotte M Deane. Cov-abdab: the coronavirus antibody database. Bioinformatics, 37(5):734–735, 2021

    2021

  56. [65]

    Minigpt-4: Enhancing vision-language understanding with advanced large lan- guage models

    Deyao Zhu, Jun Chen, Xiaoqian Shen, Xiang Li, and Mohamed Elhoseiny. Minigpt-4: Enhancing vision-language understanding with advanced large lan- guage models. arXiv preprint arXiv:2304.10592, 2023

    2023 arXiv

  57. [66]

    mixup: Beyond empirical risk minimization

    Hongyi Zhang, Moustapha Cisse, Yann N Dauphin, and David Lopez-Paz. mixup: Beyond empirical risk minimization. arXiv preprint arXiv:1710.09412, 2017

    2017 arXiv

  58. [67]

    Multilayer perceptrons for classification and regression

    Fionn Murtagh. Multilayer perceptrons for classification and regression. Neurocomputing, 2(5-6):183–197, 1991

    1991

  59. [68]

    Gradient- based learning applied to document recognition

    Yann LeCun, L ´eon Bottou, Yoshua Bengio, and Patrick Haffner. Gradient- based learning applied to document recognition. Proceedings of the IEEE , 86(11):2278–2324, 1998

    1998

  60. [69]

    Deep learning for 3d point clouds: A survey

    Yulan Guo, Hanyun Wang, Qingyong Hu, Hao Liu, Li Liu, and Mohammed Bennamoun. Deep learning for 3d point clouds: A survey. IEEE Transactions on Pattern Analysis and Machine Intelligence, 43(12):4338–4364, 2020

    2020

  61. [70]

    Deep learning for object detection and scene perception in self-driving cars: Survey, challenges, and open issues

    Abhishek Gupta, Alagan Anpalagan, Ling Guan, and Ahmed Shaharyar Khwaja. Deep learning for object detection and scene perception in self-driving cars: Survey, challenges, and open issues. Array, 10:100057, 2021

    2021

  62. [72]

    Deep learning enables therapeutic antibody optimization in mammalian cells by deciphering high-dimensional protein sequence space

    Derek M Mason, Simon Friedensohn, C ´edric R Weber, Christian Jordi, Bas- tian Wagner, Simon Meng, Pablo Gainza, Bruno E Correia, and Sai T Reddy. Deep learning enables therapeutic antibody optimization in mammalian cells by deciphering high-dimensional protein sequence space....

    2019

  63. [73]

    Optimization of therapeutic antibodies by predicting antigen speci- ficity from antibody sequence via deep learning.Nature Biomedical Engineering, 5(6):600–612, 2021

    Derek M Mason, Simon Friedensohn, C ´edric R Weber, Christian Jordi, Bas- tian Wagner, Simon M Meng, Roy A Ehling, Lucia Bonati, Jan Dahinden, Pablo 52 Gainza, et al. Optimization of therapeutic antibodies by predicting antigen speci- ficity from antibody sequence via deep lea...

    2021

  64. [74]

    Predicting unseen antibodies’ neutralizability via adaptive graph neural networks

    Jie Zhang, Yishan Du, Pengfei Zhou, Jinru Ding, Shuai Xia, Qian Wang, Feiyang Chen, Mu Zhou, Xuemei Zhang, Weifeng Wang, et al. Predicting unseen antibodies’ neutralizability via adaptive graph neural networks. Nature Machine Intelligence, 4(11):964–976, 2022

    2022

  65. [75]

    A genetic approach to mammalian glycan function

    John B Lowe and Jamey D Marth. A genetic approach to mammalian glycan function. Annual Review of Biochemistry, 72(1):643–691, 2003

    2003

  66. [77]

    Lvlm-ehub: A compre- hensive evaluation benchmark for large vision-language models

    Peng Xu, Wenqi Shao, Kaipeng Zhang, Peng Gao, Shuo Liu, Meng Lei, Fanqing Meng, Siyuan Huang, Yu Qiao, and Ping Luo. Lvlm-ehub: A compre- hensive evaluation benchmark for large vision-language models. arXiv preprint arXiv:2306.09265, 2023

    2023 arXiv

  67. [78]

    Maxi- mum classifier discrepancy for unsupervised domain adaptation

    Kuniaki Saito, Kohei Watanabe, Yoshitaka Ushiku, and Tatsuya Harada. Maxi- mum classifier discrepancy for unsupervised domain adaptation. In Proceedings of the IEEE conference on computer vision and pattern recognition , pages 3723–3732, 2018

    2018

  68. [79]

    Fully hardware-implemented memristor convolutional neural network

    Peng Yao, Huaqiang Wu, Bin Gao, Jianshi Tang, Qingtian Zhang, Wenqiang Zhang, J Joshua Yang, and He Qian. Fully hardware-implemented memristor convolutional neural network. Nature, 577(7792):641–646, 2020

    2020

  69. [80]

    A novel method for fabricating self-ordered porous anodic alumina with wide interpore distance using phosphoric/oxalic acid mixed electrolyte

    Yan-fang Xu, Hao Liu, Xiao-jiu Li, Wei-min Kang, Bo-wen Cheng, and Xiao- jie Li. A novel method for fabricating self-ordered porous anodic alumina with wide interpore distance using phosphoric/oxalic acid mixed electrolyte. Materials Letters, 151:79–81, 2015

    2015

  70. [81]

    Revisiting anodic alumina templates: From fabrication to applications

    Alejandra Ruiz-Clavijo, Olga Caballero-Calero, and Marisol Mart ´ın-Gonz´alez. Revisiting anodic alumina templates: From fabrication to applications. Nanoscale, 13(4):2227–2265, 2021

    2021

  71. [82]

    A neuromorphic bionic eye with filter-free color vision using hemispherical perovskite nanowire array retina

    Zhenghao Long, Xiao Qiu, Chak Lam Jonathan Chan, Zhibo Sun, Zhengnan Yuan, Swapnadeep Poddar, Yuting Zhang, Yucheng Ding, Leilei Gu, Yu Zhou, et al. A neuromorphic bionic eye with filter-free color vision using hemispherical perovskite nanowire array retina. Nature Communicati...

    1972

  72. [83]

    All inorganic cesium lead iodide perovskite nanowires with stabilized cubic phase at room temperature and nanowire array-based photodetectors.Nano Letters, 17(8):4951–4957, 2017

    Aashir Waleed, Mohammad Mahdi Tavakoli, Leilei Gu, Shabeeb Hussain, Daquan Zhang, Swapnadeep Poddar, Ziyi Wang, Rongjun Zhang, and Zhiy- ong Fan. All inorganic cesium lead iodide perovskite nanowires with stabilized cubic phase at room temperature and nanowire array-based phot...

    2017

  73. [84]

    In-memory learning with analog resistive switching memory: A review and perspective

    Yue Xi, Bin Gao, Jianshi Tang, An Chen, Meng-Fan Chang, Xiaobo Sharon Hu, Jan Van Der Spiegel, He Qian, and Huaqiang Wu. In-memory learning with analog resistive switching memory: A review and perspective. Proceedings of the IEEE, 109(1):14–42, 2020

    2020

  74. [85]

    Amant, Amir Yazdanbakhsh, Jongse Park, Bradley Thwaites, Hadi Esmaeilzadeh, Arjang Hassibi, Luis Ceze, and Doug Burger

    Ren ´ee St. Amant, Amir Yazdanbakhsh, Jongse Park, Bradley Thwaites, Hadi Esmaeilzadeh, Arjang Hassibi, Luis Ceze, and Doug Burger. General-purpose 53 code acceleration with limited-precision analog computation. ACM SIGARCH Computer Architecture News, 42(3):505–516, 2014

    2014

  75. [87]

    Accelerator-friendly neural-network training: Learning vari- ations and defects in rram crossbar

    Lerong Chen, Jiawen Li, Yiran Chen, Qiuping Deng, Jiyuan Shen, Xiaoyao Liang, and Li Jiang. Accelerator-friendly neural-network training: Learning vari- ations and defects in rram crossbar. In Design, Automation & Test in Europe Conference & Exhibition (DATE), 2017, pages 19–2...

    2017

  76. [88]

    Multi-level switching of triple-layered taox rram with excellent reliability for storage class memory

    Seung Ryul Lee, Young-Bae Kim, Man Chang, Kyung Min Kim, Chang Bum Lee, Ji Hyun Hur, Gyeong-Su Park, Dongsoo Lee, Myoung-Jae Lee, Chang Jung Kim, U-In Chung, In-Kyeong Yoo, and Kinam Kim. Multi-level switching of triple-layered taox rram with excellent reliability for storage ...

    2012

  77. [89]

    Bayesian optimization

    Roman Garnett. Bayesian optimization. Cambridge University Press, Cam- bridge, UK, 2023

    2023

  78. [90]

    Pointpillars: Fast encoders for object detection from point clouds

    Alex H Lang, Sourabh V ora, Holger Caesar, Lubing Zhou, Jiong Yang, and Oscar Beijbom. Pointpillars: Fast encoders for object detection from point clouds. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 12697–12705, 2019

    2019

  79. [91]

    Optimization of therapeutic antibodies by predicting antigen speci- ficity from antibody sequence via deep learning.Nature Biomedical Engineering, 5(6):600–612, 2021

    Derek M Mason, Simon Friedensohn, C ´edric R Weber, Christian Jordi, Bas- tian Wagner, Simon M Meng, Roy A Ehling, Lucia Bonati, Jan Dahinden, Pablo Gainza, et al. Optimization of therapeutic antibodies by predicting antigen speci- ficity from antibody sequence via deep learni...

    2021

  80. [92]

    Using graph con- volutional neural networks to learn a representation for glycans

    Rebekka Burkholz, John Quackenbush, and Daniel Bojar. Using graph con- volutional neural networks to learn a representation for glycans. Cell Reports, 35(11), 2021

    2021

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.