Researchers Improve Zero-Shot Activity Recognition from Wearable Sensors Using Contrastive Learning
Researchers propose a framework combining contrastive training with optimized semantic prototypes to improve zero-shot human activity recognition (HAR) using inertial measurement unit (IMU) sensor data, achieving 73.2% accuracy on unseen activity classes. The work identifies the 'modality gap'—the misalignment between sensor embeddings and text-based class representations—as a training-time phenomenon that can be substantially reduced by replacing simple label-name prototypes with richer activity descriptions. The findings also highlight that overall accuracy is a misleading metric for imbalanced benchmarks, recommending macro-averaged F1 instead.
The study systematically evaluates seven configurations of inference methods and training pipelines for zero-shot learning (ZSL) applied to IMU-based HAR on the PAMAP2 dataset, using 14 seen and 4 unseen activity classes. A temporal convolutional network (TCN) trained with cross-entropy against simple label-name Sentence-BERT prototypes achieves a mean cosine similarity of only 0.30 between sensor and text embeddings, while substituting richer, discriminative activity descriptions raises this to 0.69. The best-performing configuration—contrastive training combined with inverted softmax correction—achieves 73.2% accuracy and a macro F1 of 0.583 on unseen classes, compared to 58.3% accuracy and 0.34 macro F1 for the label-name baseline. A notable secondary finding is that richer text descriptions can reduce inter-prototype separability in Sentence-BERT space, because shared biomechanical vocabulary causes the language model to compress the prototype representations closer together; however, this compression does not negate alignment benefits as long as descriptions retain sufficiently discriminative vocabulary. The authors also argue that overall accuracy is a misleading primary metric under imbalanced class distributions and recommend macro-averaged F1 as the standard for ZSL-HAR benchmarks going forward.
What's missing
The study is limited to a single dataset (PAMAP2) with a specific train/test subject split (subjects 108 and 109 held out), leaving generalizability to other HAR datasets, sensor modalities, or subject populations undemonstrated. The choice of Sentence-BERT as the language model backbone is not ablated against other embedding models, and the sensitivity of results to the quality of manually crafted activity descriptions is not fully characterized. The paper does not address computational cost or real-world deployment constraints for the proposed pipeline.
What different sources said
- arXiv cs.LGCenter
Closing the Modality Gap in Zero-Shot HAR: Contrastive Training and Separability-Optimized Prototypes on IMU Data
Related
Gut Bacteria Enzyme Found to Break Down Heat-Processed Food Compounds, Producing Novel Biogenic Amines
Researchers have discovered that an enzyme in common gut bacteria can degrade N-epsilon-carboxymethyllysine (CML), a compound formed during thermal food processing, producing previously unknown biogenic amines. The enzyme, ornithine decarboxylase SpeC from enterobacteria, acts on CML and related modified lysine derivatives through a low-level 'underground' catalytic activity. This finding suggests a previously unrecognized communication axis between thermally processed dietary compounds and gut microbial physiology, with potential implications for host health.
Full-Length Gene Sequencing Reveals Two Distinct Bacterial Communities in Black-Legged Ticks Expanding Into Canada
Researchers used Oxford Nanopore full-length 16S rRNA gene sequencing to characterize the microbiome of Ixodes scapularis black-legged ticks collected in Nova Scotia, Canada, distinguishing between tick-adapted bacteria and environmentally acquired bacteria. The study comes as I. scapularis — the primary vector of Lyme disease — is rapidly expanding northward into Canada due to climate change. The findings suggest that environmentally derived bacteria in tick microbiomes are not mere contamination, which has implications for how tick microbiome data is collected and interpreted across surveillance studies.
Study Identifies Metabolic Link Between Cell Envelope Stress and Biofilm Formation in Bacteria
Researchers have discovered that the metabolite acetyl-CoA directly inhibits enzymes that degrade the bacterial signaling molecule c-di-GMP, connecting cell envelope biosynthesis stress to biofilm formation in Pseudomonas aeruginosa. The study found that sub-inhibitory concentrations of antibiotics targeting early peptidoglycan biosynthesis — but not other antibiotic classes — elevate c-di-GMP levels by reducing phosphodiesterase activity, with acetyl-CoA competing for the enzyme active site. Because the relevant enzyme domain is broadly conserved across bacterial species, this checkpoint mechanism may be widespread and could have implications for understanding antibiotic-induced biofilm responses.