Pith. sign in

REVIEW 3 major objections 5 minor 67 references

Neuro-Photonix: Enabling Near-Sensor Neuro-Symbolic AI Computing on Silicon Photonics Substrate

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 Neuro-Photonix claims a near-sensor photonic core runs neuro-symbolic AI at 30 GOPS/W.

desk verdict A plausible first near-sensor neuro-symbolic photonic accelerator with real engineering substance, but the accuracy headline depends on a quantized PyTorch model that excludes photonic non-idealities—the paper says so itself. read the letter →

arxiv 2412.10187 v1 pith:CVCHB4R4 submitted 2024-12-13 cs.AR

Deniz Najafi , Hamza Errahmouni Barkam , Mehrdad Morsali , SungHeon Jeong , Tamoghno Das , Arman Roohi , Mahdi Nikdast , Mohsen Imani
show 1 more author
Shaahin Angizi
This is my paper · ORCID
classification cs.AR
keywords neuro-symbolicAInear-sensorcomputingsiliconphotonicsmicro-ringresonatorhyperdimensionalRAVENreasoningprocessing-in-sensorenergy-efficientaccelerator
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

Neuro-Photonix is an architecture for running neuro-symbolic AI at the sensor: it converts pixel voltages to 4-bit digital values with a comparator-based converter, drives VCSELs to encode activations as light, and performs the neural network's MAC operations in a single cycle using banks of micro-ring resonators. The same optical core is then reconfigured to multiply the network output by an encoding matrix, producing a hyperdimensional vector for symbolic reasoning. The authors report 30 GOPS/W on the neural-dynamics portion, average power reductions of 20.8x versus ASIC baselines and 4.1x versus photonic accelerators, and RAVEN reasoning accuracy within about 2 percentage points of the full-precision NVSA model. The motivation is that IoT sensor nodes cannot afford the power, ADC cost, or cloud round-trip of conventional deep-reasoning pipelines, so moving the neural and symbolic encoding work next to the pixel array could make transparent reasoning practical at the edge.

What carries the argument

The load-bearing mechanism is the photonic MAC engine in the optical core banks. Each arm of a bank contains nine micro-ring resonators, matching the common 3-by-3 convolution kernel; light from the light driver unit carries the activation, each ring attenuates the light according to the weight imprinted on it, and a photodetector at the arm's end sums the partial products to complete the multiplication-accumulate in a single cycle. Larger kernels (5x5, 7x7) and fully connected layers are handled by chaining arms and using an accumulation unit. The same structure doubles as the hypervector encoder: encoding-matrix weights from a dedicated memory are loaded into the rings, and the network output is multiplied through the banks to produce the hypervector. The comparator-based converter (CBC) is the second key piece: 15 comparators with a 4-bit output drive the light driver, eliminating the latch, encoder, and power of a conventional ADC.

What would settle it

Build or faithfully simulate the full Neuro-Photonix datapath, including the comparator-based converter, VCSEL driver, micro-ring weight tuning, photodetector summation, and 4-bit quantization, and measure end-to-end RAVEN accuracy, power, and frames/sec on the same ResNet-18/VGG9 workloads. If the measured accuracy falls more than about 2 points below the 97.99% software result at [4:4], or if the [3:4] configuration draws substantially more than 2.71 W while delivering the reported throughput, the central claim of preserving accuracy at 30 GOPS/W would be falsified.

Watch

Extended reading notes

Core claim

The paper's central claim is that one reconfigurable optical core can absorb the entire neural-and-symbolic preprocessing pipeline of a neuro-vector-symbolic model: weights are held as resonant-wavelength settings on micro-ring resonators, activations are carried by laser light intensity, a photodetector accumulates partial products to finish each MAC in one clock cycle, and the same banks are reloaded with encoding-matrix weights to turn the network output into a 1024-dimensional hypervector. On the RAVEN benchmark, the adapted NVSA implementation reaches 97.99% average accuracy versus 98.5% for the full-precision NVSA, with no more than a 2-point drop on the hardest 3x3 configuration, using quantized weights and activations. The authors also report that replacing a conventional ADC with the comparator-based converter (CBC) and using VCSELs directly as light modulators removes the dominant conversion and tuning costs, and that reusing weights across activations (RU) cuts neural-dynamics energy by roughly 800x and processing time by roughly 400x compared with retuning every cycle (NRU).

Load-bearing premise

The paper's accuracy and power claims assume that the quantized software model behaves the same on the analog photonic hardware; the reported RAVEN accuracy excludes MR tuning error, photodetector noise, crosstalk, and thermal drift, and the authors state that the main accuracy drop already comes from the ADC-less imager affecting the first layer.

Editorial extensions

If this is right

  • If the reported numbers hold, the [3:4] configuration would draw about 2.71 W and deliver about 117.65 thousand frames/s per watt on the evaluated VGG9/CIFAR-100 workload, fitting within an edge power budget.
  • Cloud transmission cost drops by roughly 128x, because only the 1024-dimensional hypervector (about 512 bytes) is sent instead of the full 65,536-byte input frame over a BLE link.
  • Reusing weights across activation strides in the RU method brings neural-dynamics processing time down by roughly 400x and symbolic-encoding time by roughly 1000x relative to always retuning the micro-rings, making MR tuning and DAC conversion the dominant remaining costs.
  • The architecture keeps RAVEN accuracy above 97% with 8-bit quantization at dimensionality 1024, but dropping to 4-bit at the same dimension costs about 3% on the center configuration, indicating a precision floor near 4 bits for this reasoning task.

Reading between the lines

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

  • The reported accuracy is obtained from a quantized PyTorch simulation, not from the analog photonic hardware; an implicit extension is to fold MR tuning error, photodetector noise, crosstalk, and thermal drift into the training loop, which the authors themselves hint at when noting the first-layer ADC-less imager causes the main accuracy drop.
  • Because the optical core already produces hypervectors, the same ring banks could plausibly be extended to perform in-situ similarity search or associative memory, removing the cloud step entirely; the authors list this as future work, and it is a natural next move.
  • The 30 GOPS/W headline is for neural dynamics only; an end-to-end efficiency measure that includes symbolic encoding, photonic tuning, and any residual ADC/DAC conversion would be lower, so the stated power reductions are best read as component-level rather than whole-system.
  • A testable consequence is that hardware noise-aware training should recover most of the first-layer loss, predicting that a fabricated chip with such training would match the software accuracy within noise tolerance; conversely, without it, the 'preserving accuracy' claim may not transfer to silicon.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

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 Neuro-Photonix, a near-sensor silicon-photonics accelerator for neuro-symbolic AI. The architecture couples a CMOS sensor array to a comparator-based converter (CBC) and light driver unit (LMU), which feed a reconfigurable optical core bank (OCB) of microring resonators. The same OCB implements the convolutional and fully connected layers of a neural network (single-cycle MAC operations) and then reconfigures to perform hyperdimensional (HD) encoding of the network output, producing a 1024-dimensional hypervector for cloud-side symbolic reasoning. The authors evaluate the design on RAVEN and other datasets using a quantized PyTorch implementation of the NVSA model for accuracy, and a combination of Cadence/SPICE, CACTI, and an in-house simulator for energy, latency, and power. They report 30 GOPS/W, accuracy within about 0.5 percentage points of NVSA on RAVEN, and power reductions of 20.8x/4.1x on average versus ASIC and photonic baselines, respectively.

Significance. The paper makes a valuable architectural contribution: it is, to my knowledge, the first to combine near-sensor photonic computation with hyperdimensional symbolic encoding in a single accelerator, and the mapping of the neural and symbolic workloads onto the same microring-resonator banks is well thought out. The bottom-up evaluation framework is a genuine strength: MR devices were fabricated and characterized, CMOS peripherals were simulated with a 45nm PDK in Cadence, CACTI was used for memory, and the authors compare against a broad set of external ASIC and photonic baselines using a unified in-house simulator. The systematic study of bit-precision and hypervector dimensionality in Section V.C is also useful. If the accuracy numbers were shown to survive hardware non-idealities, the claimed efficiency and near-sensor operation would be a meaningful step for edge neuro-symbolic AI. As it stands, the central 'preserving accuracy' claim is not yet supported at the hardware level, and several of the headline efficiency numbers are not fully auditable.

major comments (3)
  1. [Section V.A and Section V.F2] The application-level accuracy results in Table I and Table II, including the 97.99% average on RAVEN, are produced by a quantized PyTorch implementation of the NVSA model, not by the photonic hardware or its circuit-level model. Section V.A describes device-level characterization and Cadence/SPICE co-simulation, but these non-idealities (MR tuning error, photodetector noise, crosstalk, thermal drift, CBC comparator offset) are not propagated into the accuracy pipeline. Section V.F2 states that 'the main accuracy drop ... is due to the ADC-less imager affecting the first layer,' which confirms that hardware non-idealities are excluded from the reported accuracy. Since the abstract's 'while preserving accuracy' rests on these numbers, the accuracy claim is not established at the device-architecture level; at minimum, the paper must either add a hardware-noise-in-the-loop accuracy evaluation or explicitly re-label the reported accuracy as the algorithmic ceiling of the quantized model.
  2. [Section V.A and Section V.E] The energy, latency, and GOPS/W results depend on an in-house simulator whose model structure, input parameters, and validation are not specified. The paper states that the framework 'computes both execution time and energy consumption' and that comparisons in Table II were built with the same framework, but no equations, calibrated parameter values, or comparisons against measurements or an open-source simulator are provided. As a result, the quantitative claims in Section V.E and the 30 GOPS/W headline in the abstract cannot be independently verified. Please document the simulator's modeling assumptions and validate key components against published silicon results.
  3. [Abstract and Section V.F] The headline efficiency metrics are not internally consistent with the detailed results. The abstract and conclusion report 30 GOPS/W and average power-reduction factors of 20.8x (ASIC baselines) and 4.1x (photonic accelerators), but Section V.F reports 19x/28x/17.6x against Eyeriss/YodaNN/AppCip and 73x/24.68x/30.9x against baseline/HolyLight/CrossLight. The derivation of 20.8 and 4.1 is not shown, and GOPS/W is never defined or computed in the experimental section. Please clarify how these aggregate numbers are obtained and reconcile them with the per-baseline results.
minor comments (5)
  1. [Section V.D] 'blacktooth' should be 'Bluetooth', and the BLE 4.0 reference is missing ('[?]').
  2. [Section V.C] The text reports an average accuracy of 97.92% while Table I lists 97.99%; please reconcile.
  3. [Fig. 10(a)] The axis labels appear garbled ('328421', '8196'); the horizontal axis should presumably list dimensionality values 512, 1024, 2048, and 8192.
  4. [Table II] The CrossLight row has an undefined process-node marker and a power range of 84-390 W without explanation; please expand the caption or add footnotes.
  5. [Section III.B.1] The text says non-linear activation functions can be implemented optically, but the design appears to use an electronic activation function; clarify which approach is actually used in Neuro-Photonix.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity found; external benchmarks ground the accuracy and power claims, and the acknowledged hardware-noise omission is an evaluation gap, not a circular reduction.

full rationale

The paper's central claims—30 GOPS/W, roughly 20.8x and 4.1x power reduction, and 97.99% average RAVEN accuracy—are not derived from a fitted parameter or from a self-citation chain. Accuracy is benchmarked against external NVSA [60] and other published models on the external RAVEN dataset, and the model is a quantized PyTorch implementation whose weights are 'extracted, quantized, fine-tuned, scaled... and mapped into the OCB' (Section V-A). The energy and power comparisons use external baselines such as Eyeriss, YodaNN, AppCip, LightBulb, HolyLight, Robin, and CrossLight. Several references are self-citations (e.g., AppCip [22], OISA [39], Lightator [41]), but they are prior-art and baseline material, not load-bearing premises or uniqueness theorems. The paper itself flags the main unmodeled effect: 'The main accuracy drop in Neuro-Photonix is due to the ADC-less imager affecting the first layer, which can be improved with hardware noise-aware training' (Section V-F2). That is an omission of hardware non-idealities from the reported accuracy—a correctness or validation risk—not a case where the prediction reduces by construction to its inputs. No equation or fitted input is renamed as a prediction, and no external result is replaced by an author-imported uniqueness claim. Therefore the derivation chain is not circular.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The design relies on standard photonic and analog-circuit assumptions plus several hand-chosen sizing parameters such as D, bit precision, and bank counts. No new physical entity is postulated. The main uncharged cost is the assumption that analog photonic MAC non-idealities do not degrade the reported accuracy.

free parameters (4)
  • Hypervector dimensionality D = 1024
    Chosen after an accuracy sweep: Figure 10(a) shows 512 reduces accuracy by about 6% at full precision and 1024 matches near-SOTA, so 1024 is fit to the RAVEN accuracy target.
  • Input and weight bit precision = [3:4] and [4:4]
    Selected from a quantization sweep that balances accuracy and power; Table II reports [4:4], [3:4], and [2:4] with accuracy and power trade-offs.
  • Number of comparators in the CBC = 15 (4-bit)
    Converts analog pixel voltage into a 4-bit thermometer code; the bit-width is a design choice tied to the accuracy and energy trade-off.
  • OCB bank and MR count = 96 banks x 54 MRs = 5184 MRs
    Chosen so that 7x7 kernels and ResNet18-style layers fit in one or a few cycles; this is a hardware sizing parameter, not a physics constant.
assumptions (5)
  • domain assumption Photonic MAC with microrings and photodetectors preserves enough analog precision for 3 to 4 bit inference.
    Invoked in Sections III-B and IV-A; no measured noise, crosstalk, or thermal model is used in the accuracy evaluation.
  • domain assumption VCSEL output intensity is a monotonic and repeatable function of drive current controlled by the LDU transistors.
    Section III-B LMU description relies on this; no device characterization data are provided.
  • domain assumption The comparator-based converter encodes pixel voltages to 4-bit values with negligible error beyond quantization.
    Section III-B(a) describes the CBC; simulated waveforms are shown, but comparator offset and mismatch are not analyzed.
  • domain assumption Power and latency models from Cadence Spectre, SPICE, CACTI, and the in-house simulator are representative of a real photonic process.
    Section V-A describes the framework; no cross-validation against measurements on the full optical core is shown.
  • domain assumption Quantized PyTorch accuracy on RAVEN and CIFAR transfers to the analog optical hardware.
    Sections V-B and V-F report accuracy from a quantized model; hardware non-idealities are excluded from those numbers, as acknowledged in the Table II discussion.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Neuro-Photonix: Enabling Near-Sensor Neuro-Symbolic AI Computing on Silicon Photonics Substrate." pith.science (2026). https://pith.science/paper/CVCHB4R4

@misc{pith2026241210187,
  author       = {Pith},
  title        = {Pith review of: Neuro-Photonix: Enabling Near-Sensor Neuro-Symbolic AI Computing on Silicon Photonics Substrate},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CVCHB4R4}},
  note         = {Machine review of arXiv:2412.10187}
}
read the original abstract

Neuro-symbolic Artificial Intelligence (AI) models, blending neural networks with symbolic AI, have facilitated transparent reasoning and context understanding without the need for explicit rule-based programming. However, implementing such models in the Internet of Things (IoT) sensor nodes presents hurdles due to computational constraints and intricacies. In this work, for the first time, we propose a near-sensor neuro-symbolic AI computing accelerator named Neuro-Photonix for vision applications. Neuro-photonix processes neural dynamic computations on analog data while inherently supporting granularity-controllable convolution operations through the efficient use of photonic devices. Additionally, the creation of an innovative, low-cost ADC that works seamlessly with photonic technology removes the necessity for costly ADCs. Moreover, Neuro-Photonix facilitates the generation of HyperDimensional (HD) vectors for HD-based symbolic AI computing. This approach allows the proposed design to substantially diminish the energy consumption and latency of conversion, transmission, and processing within the established cloud-centric architecture and recently designed accelerators. Our device-to-architecture results show that Neuro-Photonix achieves 30 GOPS/W and reduces power consumption by a factor of 20.8 and 4.1 on average on neural dynamics compared to ASIC baselines and photonic accelerators while preserving accuracy.

Figures

Figures reproduced from arXiv: 2412.10187 by the authors.

Figure 1
Figure 1. MR input and through ports’ spectra after imprinting a parameter [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Neuro-symbolic AI framework consisting of neural dynamic and [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. High-level operational flow of the proposed Neuro-Photonix architec [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (9 more)
Figure 4
Figure 4. Figure 4: Neuro-Photonix architecture consisting of a sensor array and the optical core. [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Schematic of (a) Comparator-based Convertor (CBC) and (b) Light [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 7
Figure 7. Figure 7: Hyper-vector generation in OCB. B. Photonics-Friendly Encoder To achieve the conversion of output data from the DNN output into HV with acceptable efficiency, the encoding HV matrix should have a dimension of 1024. This dimension is adequate to ensure the accuracy disc…
Figure 9
Figure 9. Figure 9: Example from the RAVEN dataset [PITH_FULL_IMAGE:figures/full_fig_p007_9.png]
Figure 10
Figure 10. Figure 10: (a) Heatmap displaying accuracy across various dimensions and [PITH_FULL_IMAGE:figures/full_fig_p008_10.png]
Figure 11
Figure 11. Figure 11: Break-down of energy consumption for ResNET18 layers and [PITH_FULL_IMAGE:figures/full_fig_p009_11.png]
Figure 12
Figure 12. Figure 12: Break-down of energy consumption for ResNET18 layers and [PITH_FULL_IMAGE:figures/full_fig_p009_12.png]
Figure 14
Figure 14. Figure 14: Execution time of ResNET18 layers and encoder layer under different [PITH_FULL_IMAGE:figures/full_fig_p010_14.png]
Figure 15
Figure 15. Figure 15: Distribution of energy consumption (a) and (c) and execution time [PITH_FULL_IMAGE:figures/full_fig_p010_15.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

67 extracted references · 47 canonical work pages

  1. [1]

    Pervasive ai for iot applications: A survey on resource-efficient distributed artificial intelligence,

    E. Baccour, N. Mhaisen, A. A. Abdellatif, A. Erbad, A. Mohamed, M. Hamdi, and M. Guizani, “Pervasive ai for iot applications: A survey on resource-efficient distributed artificial intelligence,” IEEE Communications Surveys & Tutorials , vol. 24, no. 4, pp. 2366–2418, 2022

    2022

  2. [2]

    Tizbin: A low-power image sensor with event and object detection using efficient processing-in-pixel schemes,

    S. Tabrizchi, S. Angizi, and A. Roohi, “Tizbin: A low-power image sensor with event and object detection using efficient processing-in-pixel schemes,” in 2022 IEEE 40th International Conference on Computer Design (ICCD). IEEE, 2022, pp. 770–777

    work page 2022

  3. [3]

    A reconfigurable convolution-in-pixel cmos image sensor architecture,

    R. Song, K. Huang, Z. Wang, and H. Shen, “A reconfigurable convolution-in-pixel cmos image sensor architecture,” IEEE Transac- tions on Circuits and Systems for Video Technology , vol. 32, no. 10, pp. 7212–7225, 2022

    work page 2022

  4. [4]

    Macsen: A processing-in-sensor architecture integrating mac operations into image sensor for ultra-low-power bnn-based intelligent visual perception,

    H. Xu, Z. Li, N. Lin, Q. Wei, F. Qiao, X. Yin, and H. Yang, “Macsen: A processing-in-sensor architecture integrating mac operations into image sensor for ultra-low-power bnn-based intelligent visual perception,” IEEE Transactions on Circuits and Systems II: Express Briefs , vol. 68, no. 2, pp. 627–631, 2021

    work page 2021

  5. [5]

    Pisa: A non- volatile processing-in-sensor accelerator for imaging systems,

    S. Angizi, S. Tabrizchi, D. Z. Pan, and A. Roohi, “Pisa: A non- volatile processing-in-sensor accelerator for imaging systems,” IEEE Transactions on Emerging Topics in Computing, vol. 11, no. 4, pp. 962– 972, 2023

    work page 2023

  6. [6]

    A survey of deep learning for mathematical reasoning,

    P. Lu, L. Qiu, W. Yu, S. Welleck, and K.-W. Chang, “A survey of deep learning for mathematical reasoning,” arXiv preprint arXiv:2212.10535, 2022

    arXiv 2022

  7. [7]

    Neuro-symbolic computing: Advancements and challenges in hardware–software co-design,

    X. Yang, Z. Wang, X. S. Hu, C. H. Kim, S. Yu, M. Pajic, R. Manohar, Y . Chen, and H. H. Li, “Neuro-symbolic computing: Advancements and challenges in hardware–software co-design,” IEEE Transactions on Circuits and Systems II: Express Briefs , vol. 71, no. 3, pp. 1683–1689, 2024

    work page 2024

  8. [8]

    Neurosymbolic ai: The 3 rd wave,

    A. d. Garcez and L. C. Lamb, “Neurosymbolic ai: The 3 rd wave,” Artificial Intelligence Review, vol. 56, no. 11, pp. 12 387–12 406, 2023

    2023

Show all 67 references
  1. [9]

    H3dfact: Heterogeneous 3d integrated cim for factorization with holographic perceptual representations,

    Z. Wan, C.-K. Liu, M. Ibrahim, H. Yang, S. Spetalnick, T. Krishna, and A. Raychowdhury, “H3dfact: Heterogeneous 3d integrated cim for factorization with holographic perceptual representations,” in 2024 Design, Automation & Test in Europe Conference & Exhibition (DATE). IEEE, 2...

    2024

  2. [10]

    Weakly supervised neural symbolic learning for cognitive tasks,

    J. Tian, Y . Li, W. Chen, L. Xiao, H. He, and Y . Jin, “Weakly supervised neural symbolic learning for cognitive tasks,” inProceedings of the AAAI Conference on Artificial Intelligence , vol. 36, no. 5, 2022, pp. 5888– 5896

    2022

  3. [11]

    The neuro- symbolic concept learner: Interpreting scenes, words, and sentences from natural supervision,

    J. Mao, C. Gan, P. Kohli, J. B. Tenenbaum, and J. Wu, “The neuro- symbolic concept learner: Interpreting scenes, words, and sentences from natural supervision,” arXiv preprint arXiv:1904.12584 , 2019

    1904 arXiv

  4. [12]

    Probabilistic logic neural networks for reasoning,

    M. Qu and J. Tang, “Probabilistic logic neural networks for reasoning,” Advances in neural information processing systems , vol. 32, 2019

    2019

  5. [13]

    Hybrid magneto-electric fet-cmos integrated memory design for instant-on computing,

    D. Najafi, S. Tabrizchi, R. Zhou, M. Amel Solouki, A. Marshal, A. Roohi, and S. Angizi, “Hybrid magneto-electric fet-cmos integrated memory design for instant-on computing,” in Proceedings of the Great Lakes Symposium on VLSI 2024 , 2024, pp. 770–775

    2024

  6. [14]

    A 100,000 fps vision sensor with embedded 535gops/w 256 × 256 simd processor array,

    S. J. Carey, A. Lopich, D. R. Barr, B. Wang, and P. Dudek, “A 100,000 fps vision sensor with embedded 535gops/w 256 × 256 simd processor array,” in 2013 Symposium on VLSI Circuits . IEEE, 2013, pp. C182– C183

    2013

  7. [15]

    Enabling normally-off in situ computing with a magneto- electric fet-based sram design,

    D. Najafi, M. Morsali, R. Zhou, A. Roohi, A. Marshall, D. Misra, and S. Angizi, “Enabling normally-off in situ computing with a magneto- electric fet-based sram design,” IEEE Transactions on Electron Devices, vol. 71, no. 4, pp. 2742–2748, 2024

    2024

  8. [16]

    A 0.5-v real-time computational cmos image sensor with programmable kernel for feature extraction,

    T.-H. Hsu, Y .-R. Chen, R.-S. Liu, C.-C. Lo, K.-T. Tang, M.-F. Chang, and C.-C. Hsieh, “A 0.5-v real-time computational cmos image sensor with programmable kernel for feature extraction,”IEEE Journal of Solid- State Circuits, vol. 56, no. 5, pp. 1588–1596, 2020

    2020

  9. [17]

    4.9 a 1ms high-speed vision chip with 3d-stacked 140gops column-parallel pes for spatio-temporal image processing,

    T. Yamazaki, H. Katayama, S. Uehara, A. Nose, M. Kobayashi, S. Shida, M. Odahara, K. Takamiya, Y . Hisamatsu, S. Matsumoto et al. , “4.9 a 1ms high-speed vision chip with 3d-stacked 140gops column-parallel pes for spatio-temporal image processing,” in 2017 IEEE International S...

    2017

  10. [18]

    Mrima: An mram-based in-memory accelerator,

    S. Angizi et al., “Mrima: An mram-based in-memory accelerator,” IEEE TCAD, vol. 39, no. 5, pp. 1123–1136, 2019

    2019

  11. [19]

    Nese: Near-sensor event-driven scheme for low power energy harvesting sensors,

    S. Tabrizchi, M. Morsali, S. Angizi, and A. Roohi, “Nese: Near-sensor event-driven scheme for low power energy harvesting sensors,” in 2023 IEEE International Symposium on Circuits and Systems (ISCAS). IEEE, 2023, pp. 1–4

    2023

  12. [20]

    Cmp-pim: an energy-efficient comparator-based processing-in-memory neural network accelerator,

    S. Angizi, Z. He, A. S. Rakin, and D. Fan, “Cmp-pim: an energy-efficient comparator-based processing-in-memory neural network accelerator,” in Proceedings of the 55th Annual Design Automation Conference , 2018, pp. 1–6

    2018

  13. [21]

    Senputing: An ultra-low-power always-on vision perception chip featuring the deep fusion of sensing and computing,

    H. Xu, N. Lin, L. Luo, Q. Wei, R. Wang, C. Zhuo, X. Yin, F. Qiao, and H. Yang, “Senputing: An ultra-low-power always-on vision perception chip featuring the deep fusion of sensing and computing,” IEEE Trans- actions on Circuits and Systems I: Regular Papers , vol. 69, no. 1, p...

    2021

  14. [22]

    Appcip: Energy- efficient approximate convolution-in-pixel scheme for neural network acceleration,

    S. Tabrizchi, A. Nezhadi, S. Angizi, and A. Roohi, “Appcip: Energy- efficient approximate convolution-in-pixel scheme for neural network acceleration,” IEEE Journal on Emerging and Selected Topics in Circuits and Systems, vol. 13, no. 1, pp. 225–236, 2023

    2023

  15. [23]

    Apris: Approximate processing reram in-sensor architecture enabling artificial-intelligence-powered edge,

    S. Tabrizchi, R. Gaire, M. Morsali, M. Liehr, N. Cady, S. Angizi, and A. Roohi, “Apris: Approximate processing reram in-sensor architecture enabling artificial-intelligence-powered edge,” IEEE Transactions on Emerging Topics in Computing , 2024

    2024

  16. [24]

    Vitsen: Bridging vision transformers and edge computing with ad- vanced in/near-sensor processing,

    S. Tabrizchi, B. C. Reidy, D. Najafi, S. Angizi, R. Zand, and A. Roohi, “Vitsen: Bridging vision transformers and edge computing with ad- vanced in/near-sensor processing,” IEEE Embedded Systems Letters , vol. 16, no. 4, pp. 341–344, 2024

    2024

  17. [25]

    Hypersense: Hy- perdimensional intelligent sensing for energy-efficient sparse data pro- cessing,

    S. Yun, H. Chen, R. Masukawa, H. Errahmouni Barkam, A. Ding, W. Huang, A. Rezvani, S. Angizi, and M. Imani, “Hypersense: Hy- perdimensional intelligent sensing for energy-efficient sparse data pro- cessing,” Advanced Intelligent Systems , p. 2400228, 2024

    2024

  18. [26]

    Design and evaluation of a near-sensor magneto-electric fet-based event detector,

    M. Morsali, S. Tabrizchi, A. Marshall, A. Roohi, D. Misra, and S. An- gizi, “Design and evaluation of a near-sensor magneto-electric fet-based event detector,” IEEE Transactions on Electron Devices , 2023. 12

    2023

  19. [27]

    Pixel-level processing: why, what, and how?

    A. El Gamal, D. X. Yang, and B. A. Fowler, “Pixel-level processing: why, what, and how?” in Sensors, Cameras, and Applications for Digital Photography, vol. 3650. SPIE, 1999, pp. 2–13

    1999

  20. [28]

    Ressen: Imager privacy enhancement through residue arithmetic processing in sensors,

    N. Taheri, S. Tabrizchi, D. Najafi, S. Angizi, and A. Roohi, “Ressen: Imager privacy enhancement through residue arithmetic processing in sensors,” in 2024 IEEE Computer Society Annual Symposium on VLSI (ISVLSI). IEEE, 2024, pp. 349–354

    2024

  21. [29]

    Energy-efficient near-sensor event detector based on multilevel ga 2 o 3 rram,

    M. Morsali, S. Tabrizchi, R. T. Velpula, M. B. S. Muthu, H. P. T. Nguyen, M. Imani, A. Roohi, and S. Angizi, “Energy-efficient near-sensor event detector based on multilevel ga 2 o 3 rram,” in 2024 IEEE Computer Society Annual Symposium on VLSI (ISVLSI) . IEEE, 2024, pp. 331– 336

    2024

  22. [30]

    Mr-pipa: An integrated multi-level rram (hfo x) based processing-in-pixel accelerator,

    M. Abedin et al., “Mr-pipa: An integrated multi-level rram (hfo x) based processing-in-pixel accelerator,” IEEE JXCDC, 2022

    2022

  23. [31]

    Leca: In-sensor learned compressive acquisition for efficient machine vision on the edge,

    T. Ma, A. J. Boloor, X. Yang, W. Cao, P. Williams, N. Sun, A. Chakrabarti, and X. Zhang, “Leca: In-sensor learned compressive acquisition for efficient machine vision on the edge,” in Proceedings of the 50th Annual International Symposium on Computer Architecture , 2023, pp. 1–14

    2023

  24. [32]

    Hirise: High-resolution image scaling for edge ml via in- sensor compression and selective roi,

    B. Reidy, S. Tabrizchi, M. Mohammadi, S. Angizi, A. Roohi, and R. Zand, “Hirise: High-resolution image scaling for edge ml via in- sensor compression and selective roi,” in Proceedings of the 61st ACM/IEEE Design Automation Conference , 2024, pp. 1–6

    2024

  25. [33]

    Deep mapper: A multi-channel single-cycle near-sensor dnn accelerator,

    M. Morsali, S. Tabrizchi, M. Liehr, N. Cady, M. Imani, A. Roohi, and S. Angizi, “Deep mapper: A multi-channel single-cycle near-sensor dnn accelerator,” in 2023 IEEE International Conference on Rebooting Computing (ICRC). IEEE, 2023, pp. 1–5

    2023

  26. [34]

    An energy/illumination-adaptive cmos image sensor with reconfigurable modes of operations,

    J. Choi, S. Park, J. Cho, and E. Yoon, “An energy/illumination-adaptive cmos image sensor with reconfigurable modes of operations,” IEEE Journal of Solid-State Circuits , vol. 50, no. 6, pp. 1438–1450, 2015

    2015

  27. [35]

    Ai edge devices using computing- in-memory and processing-in-sensor: from system to device,

    T.-H. Hsu, Y .-C. Chiu, W.-C. Wei, Y .-C. Lo, C.-C. Lo, R.-S. Liu, K.-T. Tang, M.-F. Chang, and C.-C. Hsieh, “Ai edge devices using computing- in-memory and processing-in-sensor: from system to device,” in 2019 IEEE International Electron Devices Meeting (IEDM) . IEEE, 2019, pp. 22–5

    2019

  28. [36]

    Crosslight: A cross- layer optimized silicon photonic neural network accelerator,

    F. Sunny, A. Mirza, M. Nikdast, and S. Pasricha, “Crosslight: A cross- layer optimized silicon photonic neural network accelerator,” in 2021 58th ACM/IEEE Design Automation Conference (DAC) . IEEE, 2021, pp. 1069–1074

    2021

  29. [37]

    Redeye: analog convnet image sensor architecture for continuous mobile vision,

    R. LiKamWa, Y . Hou, J. Gao, M. Polansky, and L. Zhong, “Redeye: analog convnet image sensor architecture for continuous mobile vision,” ACM SIGARCH Computer Architecture News , vol. 44, no. 3, pp. 255– 266, 2016

    2016

  30. [38]

    Robin: A robust optical binary neural network accelerator,

    F. P. Sunny, A. Mirza, M. Nikdast, and S. Pasricha, “Robin: A robust optical binary neural network accelerator,” ACM Transactions on Em- bedded Computing Systems (TECS) , vol. 20, no. 5s, pp. 1–24, 2021

    2021

  31. [39]

    Oisa: Architecting an optical in-sensor accelerator for efficient visual computing,

    M. Morsali, S. Tabrizchi, D. Najafi, M. Imani, M. Nikdast, A. Roohi, and S. Angizi, “Oisa: Architecting an optical in-sensor accelerator for efficient visual computing,” in 2024 Design, Automation & Test in Europe Conference & Exhibition (DATE) . IEEE, 2024, pp. 1–6

    2024

  32. [40]

    Silicon photonics codesign for deep learning,

    Q. Cheng, J. Kwon, M. Glick, M. Bahadori, L. P. Carloni, and K. Bergman, “Silicon photonics codesign for deep learning,” Proceed- ings of the IEEE , vol. 108, no. 8, pp. 1261–1282, 2020

    2020

  33. [41]

    Lightator: An optical near-sensor accelerator with compressive acquisition enabling versatile image pro- cessing,

    M. Morsali, B. Reidy, D. Najafi, S. Tabrizchi, M. Imani, M. Nikdast, A. Roohi, R. Zand, and S. Angizi, “Lightator: An optical near-sensor accelerator with compressive acquisition enabling versatile image pro- cessing,” in Proceedings of the 61st ACM/IEEE Design Automation Conf...

    2024

  34. [42]

    A new optical information processing device design for internet of brain things application,

    C. Wang, Y . Wang, and H. Li, “A new optical information processing device design for internet of brain things application,” IEEE Access , vol. 6, pp. 75 052–75 061, 2018

    2018

  35. [43]

    Holylight: A nanophotonic accelerator for deep learning in data centers,

    W. Liu, W. Liu, Y . Ye, Q. Lou, Y . Xie, and L. Jiang, “Holylight: A nanophotonic accelerator for deep learning in data centers,” in 2019 Design, Automation & Test in Europe Conference & Exhibition (DATE). IEEE, 2019, pp. 1483–1488

    2019

  36. [44]

    Neuro-symbolic ai approaches for sensor-based human activity recognition,

    L. Arrotta et al., “Neuro-symbolic ai approaches for sensor-based human activity recognition,” 2024

    2024

  37. [45]

    Raven: A dataset for relational and analogical visual reasoning,

    C. Zhang, F. Gao, B. Jia, Y . Zhu, and S.-C. Zhu, “Raven: A dataset for relational and analogical visual reasoning,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2019, pp. 5317–5327

    2019

  38. [46]

    Arxon: A framework for approximate communication over photonic networks- on-chip,

    F. P. Sunny, A. Mirza, I. Thakkar, M. Nikdast, and S. Pasricha, “Arxon: A framework for approximate communication over photonic networks- on-chip,” IEEE Transactions on Very Large Scale Integration (VLSI) Systems, vol. 29, no. 6, pp. 1206–1219, 2021

    2021

  39. [47]

    Light- bulb: A photonic-nonvolatile-memory-based accelerator for binarized convolutional neural networks,

    F. Zokaee, Q. Lou, N. Youngblood, W. Liu, Y . Xie, and L. Jiang, “Light- bulb: A photonic-nonvolatile-memory-based accelerator for binarized convolutional neural networks,” in 2020 Design, Automation & Test in Europe Conference & Exhibition (DATE). IEEE, 2020, pp. 1438–1443

    2020

  40. [48]

    Hardware-software co-design of slimmed optical neural networks,

    Z. Zhao, D. Liu, M. Li, Z. Ying, L. Zhang, B. Xu, B. Yu, R. T. Chen, and D. Z. Pan, “Hardware-software co-design of slimmed optical neural networks,” in Proceedings of the 24th Asia and South Pacific Design Automation Conference, 2019, pp. 705–710

    2019

  41. [49]

    Silicon microring resonators,

    W. Bogaerts, P. De Heyn, T. Van Vaerenbergh, K. De V os, S. K. Selvaraja, T. Claes, P. Dumon, P. Bienstman, D. Van Thourhout, and R. Baets, “Silicon microring resonators,” Laser & Photonics Reviews , pp. 47–73, 2012

    2012

  42. [50]

    A silicon photonic accelerator for convolutional neural networks with heterogeneous quantization,

    F. Sunny, M. Nikdast, and S. Pasricha, “A silicon photonic accelerator for convolutional neural networks with heterogeneous quantization,” in Proceedings of the Great Lakes Symposium on VLSI 2022 , 2022, pp. 367–371

    2022

  43. [51]

    Classification using hyperdimensional comput- ing: A review,

    L. Ge and K. K. Parhi, “Classification using hyperdimensional comput- ing: A review,” IEEE Circuits and Systems Magazine, vol. 20, no. 2, pp. 30–47, 2020

    2020

  44. [52]

    Assessing robustness of hyperdimensional computing against errors in associative memory,

    S. Zhang, R. Wang, J. J. Zhang, A. Rahimi, and X. Jiao, “Assessing robustness of hyperdimensional computing against errors in associative memory,” in 2021 IEEE 32nd International Conference on Application- specific Systems, Architectures and Processors (ASAP) . IEEE, 2021, pp. 211–217

    2021

  45. [53]

    In-memory hyperdimensional computing,

    G. Karunaratne, M. Le Gallo, G. Cherubini, L. Benini, A. Rahimi, and A. Sebastian, “In-memory hyperdimensional computing,” Nature Electronics, vol. 3, no. 6, pp. 327–337, 2020

    2020

  46. [54]

    Hyperdimensional computing vs. neural networks: Comparing architecture and learning process,

    D. Ma, C. Hao, and X. Jiao, “Hyperdimensional computing vs. neural networks: Comparing architecture and learning process,” in 2024 25th International Symposium on Quality Electronic Design (ISQED). IEEE, 2024, pp. 1–5

    2024

  47. [55]

    High- density magnetic flash adc using domain-wall motion and pre-charge sense amplifiers,

    Y . K. Upadhyaya, M. K. Gupta, M. Hasan, and S. Maheshwari, “High- density magnetic flash adc using domain-wall motion and pre-charge sense amplifiers,” IEEE Transactions on Magnetics , vol. 52, no. 6, pp. 1–10, 2016

    2016

  48. [56]

    Tron: Transformer neural network acceleration with non-coherent silicon photonics,

    S. Afifi, F. Sunny, M. Nikdast, and S. Pasricha, “Tron: Transformer neural network acceleration with non-coherent silicon photonics,” in Proceedings of the Great Lakes Symposium on VLSI 2023 , 2023, pp. 15–21

    2023

  49. [57]

    Reprogrammable electro-optic nonlinear activation functions for optical neural networks,

    I. A. Williamson, T. W. Hughes, M. Minkov, B. Bartlett, S. Pai, and S. Fan, “Reprogrammable electro-optic nonlinear activation functions for optical neural networks,” IEEE Journal of Selected Topics in Quantum Electronics, vol. 26, no. 1, pp. 1–12, 2019

    2019

  50. [58]

    [Online]

    (2011) Ncsu eda freepdk45. [Online]. Available: http://www.eda.ncsu. edu/wiki/FreePDK45:Contents

    2011

  51. [59]

    Cacti 5.1 technical report,

    S. Thoziyoor, N. Muralimanohar, J. H. Ahn, and N. P. Jouppi, “Cacti 5.1 technical report,” HP Laboratories, Palo Alto , 2008

    2008

  52. [60]

    A neuro-vector-symbolic architecture for solving raven’s progressive matrices,

    M. Hersche, M. Zeqiri, L. Benini, A. Sebastian, and A. Rahimi, “A neuro-vector-symbolic architecture for solving raven’s progressive matrices,” Nature Machine Intelligence, vol. 5, no. 4, pp. 363–375, 2023

    2023

  53. [61]

    Effective abstract reasoning with dual- contrast network,

    T. Zhuo and M. Kankanhalli, “Effective abstract reasoning with dual- contrast network,” 2021, presented at the International Conference on Learning Representations (ICLR)

    2021

  54. [62]

    Abstract spatial-temporal reasoning via probabilistic abduction and execution,

    C. Zhang, B. Jia, S.-C. Zhu, and Y . Zhu, “Abstract spatial-temporal reasoning via probabilistic abduction and execution,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021, pp. 9736–9746

    2021

  55. [63]

    Scale-localized abstract reasoning,

    Y . Benny, N. Pekar, and L. Wolf, “Scale-localized abstract reasoning,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021, pp. 12 557–12 565

    2021

  56. [64]

    The scattering compositional learner: Discovering objects, attributes, relationships in analogical rea- soning,

    Y . Wu, H. Dong, R. Grosse, and J. Ba, “The scattering compositional learner: Discovering objects, attributes, relationships in analogical rea- soning,” 2020

    2020

  57. [65]

    Random features for large-scale kernel machines,

    A. Rahimi and B. Recht, “Random features for large-scale kernel machines,” Advances in neural information processing systems , vol. 20, 2007

    2007

  58. [66]

    Eyeriss: An energy- efficient reconfigurable accelerator for deep convolutional neural net- works,

    Y .-H. Chen, T. Krishna, J. S. Emer, and V . Sze, “Eyeriss: An energy- efficient reconfigurable accelerator for deep convolutional neural net- works,” IEEE journal of solid-state circuits, vol. 52, no. 1, pp. 127–138, 2016

    2016

  59. [67]

    Yodann: An architecture for ultralow power binary-weight cnn acceleration,

    R. Andri, L. Cavigelli, D. Rossi, and L. Benini, “Yodann: An architecture for ultralow power binary-weight cnn acceleration,” IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems , vol. 37, no. 1, pp. 48–60, 2017

    2017

Pith tools

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