← Back to all news
arXiv Machine Learning By Xin Zhang, Liangxiu Han, Yue Shi, Tam Sobeih

LDAC-Net: A Learnable Multi-Lag Differencing Attention-Convolution Network for Drift-Robust Recognition with Low-Cost MOX Gas Sensors

Read the original on arXiv Machine Learning →

LDAC‑Net is a learnable multi‑lag differencing attention‑convolution network designed for drift‑robust recognition with low‑cost metal‑oxide (MOX) gas sensors. It replaces fixed first‑order temporal differencing with a front‑end that applies window‑conditioned statistical affine normalisation and learnable multi‑lag differencing, followed by an attention‑convolution backbone that captures both local transients and long‑range temporal dependencies. On the 50‑class SmellNet‑Base task, LDAC‑Net attains 68.2 % top‑1 accuracy, outperforming the best fixed‑lag model by about 14 percentage points and a raw‑input Transformer by over 30 points, and it also transfers well to other benchmarks such as SmellNet‑Mixtures and eNose‑Drift.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Machine Learning.

Related stories

arXiv Statistics ML
Oct 28

Online gas classification at the edge

The paper introduces a lightweight neural network inspired by insect olfaction that classifies raw gas sensor readings using a sparse, high‑dimensional binary representation. It operates online, handling random sequences of single gases and binary mixtures, and dynamically incorporates new samples while providing class‑level frequency estimates to differentiate familiar from novel stimuli. The architecture, built with binary weights and Hebbian local learning, outperforms a linear SVM on mixtures, matches an optimized nonlinear SVM, and was demonstrated on the TEXEL neuromorphic chip with 74.6% accuracy, 482 nJ per inference, and 27 µW static power.

By Ton Juny Pina, Hugh Greatorex, Elisabetta Chicca
roboticssafety
More like this →
arXiv Machine Learning
Aug 5

Task-Oriented Candidate-Latent Feedback for Coarse-to-Fine Sensing in Distributed OFDM-ISAC Networks

arXiv:2608. 03319v1 Announce Type: cross Abstract: Future integrated sensing and communication (ISAC) architectures separate the sensing entity (SE) that acquires measurements from the sensing function (SF) that performs inference, creating a need for compact, task-oriented feedback on the SE-SF interface.

By Shiv Shankar, Radha Krishna Ganti, J Klutto Milleth
ragefficiency
More like this →
arXiv Machine Learning
Aug 10

MAUPITI: On-Device Prototype-Based Learning on a Smart Infrared Sensor

arXiv:2608. 07192v1 Announce Type: new Abstract: Low-resolution infrared (IR) array sensors represent an interesting solution for privacy-preserving human sensing in embedded systems.

By Beatrice Alessandra Motetti, Tanguy Dugas du Villard, Matteo Risso, Alessio Burrello, Francesco Daghero, Enrico Macii, Massimo Poncino, Marco Castellano, Alfio Basile, Daniele Jahier Pagliari
efficiencysafety
More like this →
arXiv AI
Oct 31

Multi-output LSTM soft sensors for industrial anaerobic co-digestion using LSTM-beta-VAE generative data augmentation

The paper presents a novel approach to monitoring industrial anaerobic co‑digestion by combining a generative LSTM‑beta‑VAE with a multi‑output LSTM soft sensor. By synthesising physically valid sequences of unstable operation from a year of plant data, the method augments sparse datasets and enables simultaneous prediction of key stability indicators such as pH, volatile acid ratio, and biogas production. The augmented model outperforms standard algorithms and unaugmented LSTMs, offering robust early‑warning capabilities for bioprocess stability.

By Rohit Murali, Benaissa Dekhici, Tao Chen, Dongda Zhang, Michael Short
efficiency
More like this →