Pith. sign in

REVIEW 21 cited by

Ternary Weight Networks

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 1605.04711 v3 pith:GQD3YU2F submitted 2016-05-16 cs.CV

Fengfu Li , Bin Liu , Xiaoxing Wang , Bo Zhang , Junchi Yan This is my paper · ORCID
classification cs.CV
keywords twnsprecisionternarynetworksweightsachievebettercifar-10
verification ladder T0 review T1 audit T2 compute T3 formal

Signed reviews

No signed human review yet.

0 comments
abstract

We present a memory and computation efficient ternary weight networks (TWNs) - with weights constrained to +1, 0 and -1. The Euclidian distance between full (float or double) precision weights and the ternary weights along with a scaling factor is minimized in training stage. Besides, a threshold-based ternary function is optimized to get an approximated solution which can be fast and easily computed. TWNs have shown better expressive abilities than binary precision counterparts. Meanwhile, TWNs achieve up to 16$\times$ model compression rate and need fewer multiplications compared with the float32 precision counterparts. Extensive experiments on MNIST, CIFAR-10, and ImageNet datasets show that the TWNs achieve much better result than the Binary-Weight-Networks (BWNs) and the classification performance on MNIST and CIFAR-10 is very close to the full precision networks. We also verify our method on object detection task and show that TWNs significantly outperforms BWN by more than 10\% mAP on PASCAL VOC dataset. The pytorch version of source code is available at: https://github.com/Thinklab-SJTU/twns.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 21 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Histogram-Equalized Quantization for logic-gated Residual Neural Networks

    cs.LG 2025-01 conditional novelty 7.0 of 10

    HEQ sets quantization step sizes from weight quantiles, achieving competitive CIFAR-10 accuracy and enabling logic-gated residual networks on STL-10.

  2. Behavior Backdoor for Deep Learning Models

    cs.LG 2024-12 conditional novelty 7.0 of 10

    A new backdoor attack keeps a model normal until it is quantized, after which it flips to a chosen target prediction for all inputs.

  3. SQuaT: Self-Supervised Knowledge Distillation via Student-Aware Quantized Teacher Features

    cs.LG 2026-08 conditional novelty 6.0 of 10

    SQuaT projects teacher features onto the student's quantization lattice to eliminate the unattainable-target lower bound in label-free QAT distillation.

  4. Attend to Your Own Thoughts: Breaking the Barrier for Post-Training Quantization of Reasoning LLMs through the Lens of 1.58-Bit Quantization

    cs.CL 2026-08 conditional novelty 6.0 of 10

    Using the target model's own self-generated reasoning traces as post-training quantization calibration data lets 1.58-bit ternary LLMs retain math and coding performance, beating a QAT-trained baseline with far fewer tokens.

  5. ExTernD: Expanded-Rank Ternary Decomposition Ternary LLM PTQ with Accuracy Approaching Any Quantization Level

    cs.LG 2026-07 conditional novelty 6.0 of 10

    Expanding the inner rank of a ternary LLM weight factorization makes the quantization error monotonically decrease and lets effective bit-width approach bf16 arbitrarily closely.

  6. Optimizing Binary and Ternary Neural Network Inference on RRAM Crossbars using CIM-Explorer

    cs.ET 2025-05 conditional novelty 6.0 of 10

    CIM-Explorer integrates a TVM-based compiler, multiple RRAM crossbar mappings, and simulators into a design-space exploration flow for binary and ternary neural networks.

  7. Binary and Ternary Quantization Can Enhance Feature Discrimination

    cs.LG 2025-04 conditional novelty 6.0 of 10

    For two Gaussian classes with equal variance, binary and ternary threshold quantization can increase Fisher feature discrimination, and this can translate into higher classification accuracy than full-precision data i...

  8. Forget the Data and Fine-Tuning! Just Fold the Network to Compress

    cs.LG 2025-02 conditional novelty 6.0 of 10

    Model folding compresses a network by k-means clustering similar neurons across adjacent layers and repairing activation statistics without data (Fold-AR, Fold-DIR), surpassing prior data-free methods at high sparsity.

  9. BILLNET: A Binarized Conv3D-LSTM Network with Logic-gated residual architecture for hardware-efficient video inference

    cs.CV 2025-01 conditional novelty 6.0 of 10

    A fully binarized Conv3D-LSTM model for video inference runs gesture recognition on Jester with 1.01 MB weights and 6.34 GBOPs, with an 8-9 point accuracy drop versus compact full-precision baselines.

  10. Ternary Stochastic Neuron -- Implemented with a Single Strained Magnetostrictive Nanomagnet

    cond-mat.mes-hall 2024-12 conditional novelty 6.0 of 10

    A strain-engineered magnetostrictive nanomagnet can produce the three-state activation function needed for a ternary stochastic neuron, according to Landau-Lifshitz-Gilbert simulations.

  11. Learning Filter Basis for Convolutional Neural Network Compression

    cs.CV 2019-08 conditional novelty 6.0 of 10

    A split-wise filter basis decomposition unifies earlier low-rank approximations, derives an optimal split point, and compresses VGG, ResNet, DenseNet, SRResNet, and EDSR with modest accuracy loss.

  12. Differentiable Soft Quantization: Bridging Full-Precision and Low-Bit Neural Networks

    cs.CV 2019-08 conditional novelty 6.0 of 10

    A trainable soft-tanh quantizer with learned steepness and clipping improves 1-4 bit network accuracy and yields fast ARM kernels.

  13. APQF: Agentic Profiling-Guided Structured Pruning and Mixed-Precision Quantization with Adaptive Fine-Tuning

    cs.CV 2026-08 conditional novelty 5.0 of 10

    A profiling-guided, LLM-driven framework combines structured pruning and mixed-precision quantization-aware training, reporting 13-18x bit-operation reductions with modest accuracy loss on ImageNet and CIFAR-10.

  14. Multibit neural inference in a N-ary crossbar architecture

    cs.AR 2026-04 unverdicted novelty 5.0 of 10

    Simulation of 4-state MTJ crossbars achieves 94.48% MNIST accuracy for neural inference, close to 97.56% software baseline, with analysis showing quantization as primary error and an optimal number of states per cell.

  15. Latent-Space Mean-Field Theory for Deep BitNet-like Training: Constrained Gradient Flows with Smooth Quantization and STE Limits

    math.OC 2025-08 reject novelty 5.0 of 10

    The paper's main mean-field theorem for BitNet-like training is not proven: it assumes key regularity and uses an invalid distributional limit step.

  16. DFQ-ViT: Data-Free Quantization for Vision Transformers without Fine-tuning

    cs.CV 2025-07 conditional novelty 5.0 of 10

    DFQ-ViT combines easy-to-hard synthetic sample generation with activation correction to quantize vision transformers without data or fine-tuning, outperforming PSAQ-ViT and roughly matching real-data calibration.

  17. MOGNET: A Mux-residual quantized Network leveraging Online-Generated weights

    cs.LG 2025-01 conditional novelty 5.0 of 10

    A tiny-memory CNN generates part of its weights from a cellular automaton and uses multiplexer-based quantized residual links, reporting up to 1% higher accuracy than two compressed baselines at sub-2Mb sizes.

  18. A 1Mb mixed-precision quantized encoder for image classification and patch-based compression

    cs.CV 2025-01 conditional novelty 5.0 of 10

    A reconfigurable 1 Mb mixed-precision encoder performs CIFAR-10 classification at 87.5% accuracy and patch-based VGA compression at 0.25 bpp with a full-frame decoder.

  19. Memristor-Based Selective Convolutional Circuit for High-Density Salt-and-Pepper Noise Removal

    cs.AR 2024-11 reject novelty 5.0 of 10

    A memristor-based analog circuit implementing the SeConvNet denoising algorithm is shown in PSpice simulations to restore salt-and-pepper corrupted images up to 50% noise, with a leaner variant beating the ternary sof...

  20. Efficient Ternary Weight Embedding Model: Bridging Scalability and Performance

    cs.CL 2024-11 conditional novelty 4.0 of 10

    Fine-tuning embedding models with ternary linear weights and self-distillation gives large memory savings with moderate, uneven performance drops.

  21. PSDNet and DPDNet: Efficient channel expansion, Depthwise-Pointwise-Depthwise Inverted Bottleneck Block

    cs.CV 2019-09 conditional novelty 4.0 of 10

    Using depthwise convolution to expand channels yields networks with about 60 percent of MobileNetV2's parameters and comparable CIFAR accuracy.

Pith tools