Pith. sign in

REVIEW 5 cited by

Implementing Neural Networks Over-the-Air via Reconfigurable Intelligent Surfaces

Not yet reviewed by Pith; the record is open.

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 2508.01840 v1 pith:6JLRUAST submitted 2025-08-03 cs.IT eess.SPmath.IT

Implementing Neural Networks Over-the-Air via Reconfigurable Intelligent Surfaces

classification cs.IT eess.SPmath.IT
keywords trainingsystemlayeroptimizationparametersairfcapproachcentralized
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

In this paper, we investigate reconfigurable intelligent surface (RIS)-aided multiple-input-multiple-output (MIMO) OAC systems designed to emulate the fully-connected (FC) layer of a neural network (NN) via analog OAC, where the RIS and the transceivers are jointly adjusted to engineer the ambient wireless propagation environment to emulate the weights of the target FC layer. We refer to this novel computational paradigm as AirFC. We first study the case in which the precoder, combiner, and RIS phase shift matrices are jointly optimized to minimize the mismatch between the OAC system and the target FC layer. To solve this non-convex optimization problem, we propose a low-complexity alternating optimization algorithm, where semi-closed-form/closed-form solutions for all optimization variables are derived. Next, we consider training of the system parameters using two distinct learning strategies, namely centralized training and distributed training. In the centralized training approach, training is performed at either the transmitter or the receiver, whichever possesses the channel state information (CSI), and the trained parameters are provided to the other terminal. In the distributed training approach, the transmitter and receiver iteratively update their parameters through back and forth transmissions by leveraging channel reciprocity, thereby avoiding CSI acquisition and significantly reducing computational complexity. Subsequently, we extend our analysis to a multi-RIS scenario by exploiting its spatial diversity gain to enhance the system performance. Simulation results show that the AirFC system realized by the RIS-aided MIMO configuration achieves satisfactory classification accuracy.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 5 Pith papers

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

  1. Realization of a Fully Connected Neural Layer Over-the-Air through Multi-hop Amplify-and-Forward Relays

    eess.SP 2026-03 unverdicted novelty 7.0

    Multi-hop amplify-and-forward relaying enables near-perfect over-the-air realization of fully-connected neural layers via joint optimization of precoders, combiners, and relay gains under power constraints.

  2. Over-The-Air Extreme Learning Machines with XL Reception via Nonlinear Cascaded Metasurfaces

    eess.SP 2026-01 unverdicted novelty 7.0

    An XL-MIMO system with stacked intelligent metasurfaces realizes an over-the-air extreme learning machine for binary classification, matching digital model performance in the XL regime.

  3. Invisible Walls: Privacy-Preserving ISAC Empowered by Reconfigurable Intelligent Surfaces

    eess.SP 2026-01 conditional novelty 6.0

    Randomly switching between two RIS beamforming vectors per row, with a shared-key demasking step, hides sensing information from eavesdroppers without degrading ISAC performance.

  4. Universal Approximation with XL MIMO Systems: OTA Classification via Trainable Analog Combining

    eess.SP 2025-04 unverdicted novelty 6.0

    XL-MIMO systems with analog combining perform OTA classification via ELM framework achieving over 90% accuracy with few ms latency under rich fading.

  5. Stacked Intelligent Metasurface-Aided Wave-Domain Signal Processing: From Communications to Sensing and Computing

    cs.IT 2026-01 unverdicted

    A review of stacked intelligent metasurfaces as a wave-domain signal-processing platform, covering theory, prototypes, training methods, and communications/sensing/computing applications.