Polynomial Surrogate Training for Differentiable Ternary Logic Gate Networks

• URL: http://arxiv.org/abs/2603.00302v1
• Date: Fri, 27 Feb 2026 20:37:17 GMT
• Title: Polynomial Surrogate Training for Differentiable Ternary Logic Gate Networks
• Abstract summary: We introduce Polynomial Surrogate Training (PST), which represents each ternary neuron as a degree-$ (2,2)$ with 9 learnable coefficients.<n>PST trains $2$-$3times$ faster than binary DLGNs and discover true ternary gates that are functionally diverse.
• Score: 3.2249474972573555
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Differentiable logic gate networks (DLGNs) learn compact, interpretable Boolean circuits via gradient-based training, but all existing variants are restricted to the 16 two-input binary gates. Extending DLGNs to Ternary Kleene $K_3$ logic and training DTLGNs where the UNKNOWN state enables principled abstention under uncertainty is desirable. However, the support set of potential gates per neuron explodes to $19{,}683$, making the established softmax-over-gates training approach intractable. We introduce Polynomial Surrogate Training (PST), which represents each ternary neuron as a degree-$(2,2)$ polynomial with 9 learnable coefficients (a $2{,}187\times$ parameter reduction) and prove that the gap between the trained network and its discretized logic circuit is bounded by a data-independent commitment loss that vanishes at convergence. Scaling experiments from 48K to 512K neurons on CIFAR-10 demonstrate that this hardening gap contracts with overparameterization. Ternary networks train $2$-$3\times$ faster than binary DLGNs and discover true ternary gates that are functionally diverse. On synthetic and tabular tasks we find that the UNKNOWN output acts as a Bayes-optimal uncertainty proxy, enabling selective prediction in which ternary circuits surpass binary accuracy once low-confidence predictions are filtered. More broadly, PST establishes a general polynomial-surrogate methodology whose parameterization cost grows only quadratically with logic valence, opening the door to many-valued differentiable logic.

Related papers

• EI-DDLGN: Efficient Encrypted Inference with Deep Differentiable Logic Gate Networks under TFHE [0.0]
  We present EI-DDLGN, the first in-depth study of TFHE-based DDLGN inference.<n>We also introduce Model-Fixed-Wire PBS Bypass, a semantics-preserving execution strategy.<n> Evaluations across 72 depth-width configurations on MNIST, FashionMNIST, and UCI Phishing show that DDLGNs constitute an efficient alternative to arithmetic TFHE inference.
  arXiv  Detail & Related papers  (2026-09-12T01:00:19Z)
• Fault-tolerant $|\sqrt{ \mathrm{T} }\ angle$ state preparation and injection for more efficient fine-grained quantum circuit synthesis [0.9744114320491685]
  We introduce flag fault-tolerant protocols for preparing logical $|sqrtmathrmT rangle $ magic states on the 3D tetrahedral color code.<n>We find that access to $sqrtmathrmT$ reduces the average space-time cost of synthesizing Haar-random single-qubit unitaries by approximately $20$-$30$%.
  arXiv  Detail & Related papers  (2026-08-26T13:48:38Z)
• Transformers Provably Learn to Internalize Chain-of-Thought [65.41010769606844]
  Implicit Chain-of-Thought (ICoT) trains models to internalize intermediate steps within their hidden states.<n>Log-ICoT removes thinking tokens one at a time, reducing the number of stages from linear in $k$ to logarithmic.<n>Experiments on multi-layer transformers confirm the theory and visualize how reasoning is progressively absorbed into deeper layers.
  arXiv  Detail & Related papers  (2026-05-27T15:17:06Z)
• On the Stability and Realizability of Recurrent Polynomial Surrogate Ternary Logic Gate Networks [4.91491092996493]
  Recurrent Neural Networks (RNNs) can learn to predict Signal Temporal Logic (STL) verdicts online from partial trajectories.<n>Standard RNN architectures offer no structural guarantee that outputs degrade gracefully under sensor degradation.<n>We introduce the Recurrent Differentiable Ternary Logic Gate Network (R-DTLGN), a recurrent architecture that operates over Kleene's three-valued logic $-1, 0, +1$.
  arXiv  Detail & Related papers  (2026-05-23T16:30:00Z)
• Fault-Tolerant Resource Comparison of Qudit and Qubit Encodings for Diagonal Quadratic Operators [36.94429692322632]
  Finite local Hilbert-space truncations arise naturally in quantum simulations of lattice field theories and motivate qudit encodings.<n>We compare the non-Clifford cost of implementing diagonal evolutions, exemplified by $U=e-it_x2$ in a uniform field-amplitude discretization of a real scalar field, using either one logical $d$-level qudit or $n_b=eil log drceil$ logical qubits.
  arXiv  Detail & Related papers  (2026-04-29T15:25:08Z)
• Learning with Boolean threshold functions [0.013714053458441644]
  We develop a method for training neural networks on sparse data in which the values at all nodes are strictly $pm 1$.<n>Across a range of tasks, this method provides a viable and efficient foundation for learning in neural systems.
  arXiv  Detail & Related papers  (2026-02-19T16:07:25Z)
• Differentiable Logical Programming for Quantum Circuit Discovery and Optimization [0.15229257192293197]
  We introduce a neuro-symbolic framework that reframes quantum circuit design as a differentiable logic programming problem.<n>Our model represents a scaffold of potential quantum gates and parameterized operations as a set of learnable, continuous truth values''<n>We report a hardware-aware adaptation experiment on the 133-qubit IBM Torino processor.
  arXiv  Detail & Related papers  (2026-02-09T16:40:19Z)
• Efficient circuit compression by multi-qudit entangling gates in linear optical quantum computation [1.9157770789584179]
  We show the existence of multi-level control ent-Z gates for qudits encoded in multiple modes in LOQC.<n>We present explicit linear optical schemes that realize such operations.<n>Our results improve upon a key scalability limitation and significantly improve the efficiency of LOQC architectures.
  arXiv  Detail & Related papers  (2026-02-09T08:52:41Z)
• NEXUS: Bit-Exact ANN-to-SNN Equivalence via Neuromorphic Gate Circuits with Surrogate-Free Training [0.7614628596146601]
  Spiking Neural Networks (SNNs) promise energy-efficient computing through event-driven sparsity.<n>Existing approaches sacrifice accuracy by approximating continuous values with discrete spikes.<n>We propose NEXUS, a framework that achieves bit-exact ANN-to-SNN equivalence.
  arXiv  Detail & Related papers  (2026-01-29T05:23:56Z)
• Verifiable Quantum Advantage via Optimized DQI Circuits [2.149968465453488]
  Decoded Quantum Interferometry (DQI) provides a framework for superpolynomial quantum speedups.<n>We apply DQI to the Optimal Polynomial Intersection (OPI) problem, whose code is Reed-Solomon (RS)<n>We establish that DQI for OPI is the first known candidate for verifiable quantum advantage with optimal speedup.
  arXiv  Detail & Related papers  (2025-10-13T03:19:28Z)
• R-Stitch: Dynamic Trajectory Stitching for Efficient Reasoning [80.104336426172]
  Chain-of-thought (CoT) enhances problem-solving ability of large language models.<n>CoT incurs substantial inference cost due to long autoregressive trajectories.<n>We introduce R-Stitch, a training-free hybrid decoding framework.
  arXiv  Detail & Related papers  (2025-07-23T08:14:36Z)
• Reasoning by Superposition: A Theoretical Perspective on Chain of Continuous Thought [64.43689151961054]
  We prove that a two-layer transformer with $D$ steps of continuous CoTs can solve the directed graph reachability problem.<n>In our construction, each continuous thought vector is a superposition state that encodes multiple search frontiers simultaneously.
  arXiv  Detail & Related papers  (2025-05-18T18:36:53Z)
• Efficient fault-tolerant code switching via one-way transversal CNOT gates [0.0]
  We present a code scheme that respects the constraints of FT circuit design by only making use of switching gates.<n>We analyze application of the scheme to low-distance color codes, which are suitable for operation in existing quantum processors.<n>We discuss how the scheme can be implemented with a large degree of parallelization, provided that logical auxiliary qubits can be prepared reliably enough.
  arXiv  Detail & Related papers  (2024-09-20T12:54:47Z)
• Identifying good directions to escape the NTK regime and efficiently learn low-degree plus sparse polynomials [52.11466135206223]
  We show that a wide two-layer neural network can jointly use the Tangent Kernel (NTK) and the QuadNTK to fit target functions. This yields an end to end convergence and guarantee with provable sample improvement over both the NTK and QuadNTK on their own.
  arXiv  Detail & Related papers  (2022-06-08T06:06:51Z)
• GDP: Stabilized Neural Network Pruning via Gates with Differentiable Polarization [84.57695474130273]
  Gate-based or importance-based pruning methods aim to remove channels whose importance is smallest. GDP can be plugged before convolutional layers without bells and whistles, to control the on-and-off of each channel. Experiments conducted over CIFAR-10 and ImageNet datasets show that the proposed GDP achieves the state-of-the-art performance.
  arXiv  Detail & Related papers  (2021-09-06T03:17:10Z)
• Realization of arbitrary doubly-controlled quantum phase gates [62.997667081978825]
  We introduce a high-fidelity gate set inspired by a proposal for near-term quantum advantage in optimization problems. By orchestrating coherent, multi-level control over three transmon qutrits, we synthesize a family of deterministic, continuous-angle quantum phase gates acting in the natural three-qubit computational basis.
  arXiv  Detail & Related papers  (2021-08-03T17:49:09Z)
• A Biased Graph Neural Network Sampler with Near-Optimal Regret [57.70126763759996]
  Graph neural networks (GNN) have emerged as a vehicle for applying deep network architectures to graph and relational data. In this paper, we build upon existing work and treat GNN neighbor sampling as a multi-armed bandit problem. We introduce a newly-designed reward function that introduces some degree of bias designed to reduce variance and avoid unstable, possibly-unbounded payouts.
  arXiv  Detail & Related papers  (2021-03-01T15:55:58Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.