Results 91 to 100 of about 3,973,078 (288)

Feature Discriminativity Estimation in CNNs for Transfer Learning

, 2019
The purpose of feature extraction on convolutional neural networks is to reuse deep representations learnt for a pre-trained model to solve a new, potentially unrelated problem. However, raw feature extraction from all layers is unfeasible given the massive size of these networks.
Giménez Ábalos, Víctor, Vilalta Arias, Armand, Garcia Gasulla, Dario, Labarta Mancho, Jesús José, Ayguadé Parra, Eduard   +4 more
openaire   +4 more sources

Design and analysis strategies for robust microbiome ageing research

FEBS Letters, EarlyView.
The gut microbiome changes with age and associates with age‐related morbidity and mortality, establishing it as a potential biomarker and intervention target for ageing. Realising this potential requires methodological rigour, yet distinguishing biological signals from methodological artefacts remains challenging across cohorts. This review provides an
Mark Olenik, İsmail Güderer, Yi Wang, Tayyaba Alvi, Prasoon Pandey, Handan Melike Dönertaş   +5 more
wiley   +1 more source

ABL kinase‐dependent phosphorylation of SH proteins promotes their direct interaction with CRK family SH2 domains

FEBS Letters, EarlyView.
CT10 regulator of kinase (CRK) and CRK‐Like (CRKL) are signaling adaptors driving cell adhesion, motility, differentiation, and proliferation. SH2‐domain containing (SH) proteins are enriched in YXXP motifs which when phosphorylated create preferred binding sites for CRK family SH2 domains.
Phoebe M. Cousens, Brendan W. Chandler, Anna M. Schmoker, Jaye L. Weinert, Charlotte A. Kearns, Warren T. Yacawych, Amila Šemić, Alicia M. Ebert, Bryan A. Ballif   +8 more
wiley   +1 more source

Sparse Autoencoder Features for Classifications and Transferability

Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing
Sparse Autoencoders (SAEs) provide potentials for uncovering structured, human-interpretable representations in Large Language Models (LLMs), making them a crucial tool for transparent and controllable AI systems. We systematically analyze SAE for interpretable feature extraction from LLMs in safety-critical classification tasks.
Gallifant, Jack, Chen, Shan, Sasse, Kuleen, Aerts, Hugo, Hartvigsen, Thomas, Bitterman, Danielle S.   +5 more
openaire   +4 more sources

Reconstructing enzyme evolution by protein engineering

FEBS Letters, EarlyView.
Natural enzyme evolution can be retraced by protein engineering methods such as directed evolution, rational design, and ancestral sequence reconstruction. These approaches reveal how enzymes emerged from ligand‐binding scaffolds, developed varying substrate preferences, formed oligomeric complexes, adapted to environmental changes, and evolved novel ...
Lukas Drexler, Markus R. Busch, Reinhard Sterner   +2 more
wiley   +1 more source

A construction of heterogeneous transfer learning model based on associative fusion of image feature data

Scientific Reports
Source images and predicted target images differ in image features. When heterogeneous transfer learning is applied some difficulties and further issues appear. For example, noise in image recognition appears and is required to be reduced.
Wen-Fei Tian, Ming Chen, Zhong Shu, Xue-jun Tian   +3 more
doaj   +1 more source

Investigating transcription factor dynamics in health and disease using FRAP

FEBS Letters, EarlyView.
FRAP analysis of GFP‐tagged transcription factors reveals how molecular mobility and target engagement change in response to drug treatment. By combining live‐cell imaging, quantitative model fitting, and statistical analysis, this approach uncovers transcription factor dynamics linked to disease mechanisms, providing a powerful framework for ...
Kannan Govindaraj, Carolina Vazquez Garzon, Jose Joaquin Velasco, Janine N. Post   +3 more
wiley   +1 more source

Articulatory feature recognition using dynamic Bayesian networks [PDF]

, 2007
We describe a dynamic Bayesian network for articulatory feature recognition. The model is intended to be a component of a speech recognizer that avoids the problems of conventional ``beads-on-a-string'' phoneme-based models. We demonstrate that the model
Simon King, King, S.; id_orcid, Frankel, Joe, Frankel, J., Wester, M.; id_orcid, Joe Frankel, Mirjam Wester, King, Simon, Wester, Mirjam   +8 more
core   +1 more source

The role of classifiers in feature selection: Number vs nature [PDF]

, 2008
This thesis was submitted for the degree of Doctor of Philosophy and awarded by Brunel University.Wrapper feature selection approaches are widely used to select a small subset of relevant features from a dataset.
Chrysostomou, Kyriacos Andrews
core   +3 more sources

Feature discrimination learning transfers to noisy displays in complex stimuli

Frontiers in Cognition
IntroductionPerception under noisy conditions requires not only feature identification but also a process whereby target features are selected and noise is filtered out (e.g., when identifying an animal hiding in the savannah).
Orly Azulai, Lilach Shalev, Lilach Shalev, Carmel Mevorach, Carmel Mevorach   +4 more
doaj   +1 more source