multilayer perceptron
PulseAugur coverage of multilayer perceptron — every cluster mentioning multilayer perceptron across labs, papers, and developer communities, ranked by signal.
20 day(s) with sentiment data
-
New GAT-MLP model improves Maximum Clique Problem solver selection
Researchers have developed a novel framework to improve the selection of algorithms for the Maximum Clique Problem (MCP), an NP-hard computational challenge. The proposed system integrates traditional machine learning t…
-
Robotics motion feasibility prediction improved with new Transformer model
Researchers have developed a new method for predicting motion feasibility in robotics, particularly for cluttered environments. This approach uses a point-cloud-based Transformer architecture, named GRASPFC-PTX, to lear…
-
Linear models with optimized preprocessing match advanced architectures in time-series forecasting
Researchers propose that optimizing preprocessing, rather than scaling model architectures, can significantly improve time-series forecasting accuracy. Using Ridge regression as a testbed, they found that optimal lookba…
-
KANs show comparable, sometimes inferior, performance to MLPs and GNNs in aerodynamics research · 3 sources tracked
A new research paper compares the performance of Kolmogorov Arnold Networks (KANs) against Multilayer Perceptrons (MLPs) and Graph Neural Networks (GNNs) for aerodynamic prediction tasks. While KANs demonstrate good per…
-
Neural texture compression uses hypernetworks for real-time decoding
Researchers have developed a novel method for neural texture compression using hypernetworks. This approach trains a single hypernetwork to generate both latent features and the weights/biases for a Multi-Layer Perceptr…
-
New method predicts hardness for combinatorial auction problems
A research paper proposes a new approach to solving the Winner Determination Problem (WDP) in combinatorial auctions, which is known to be NP-hard. Instead of trying to replace existing solvers with graph neural network…
-
New neural architecture THEIA learns complete Kleene three-valued logic
Researchers have developed THEIA, a novel modular neural architecture comprising 2.75 million parameters, capable of learning the complete Kleene three-valued logic (K3) truth table directly from data. While Transformer…
-
Low-power analogue neural networks use trainable nonlinear connections
Researchers have developed low-power analogue neural networks that utilize trainable nonlinear connections, inspired by Kolmogorov-Arnold networks. These networks compute directly with analogue device physics, placing t…
-
New framework PeLAP-A prunes latent diffusion models, revealing 'sparsity collapse'
Researchers have introduced PeLAP-A, a framework designed to make latent diffusion models more lightweight by adaptively pruning unimportant channels in the latent space. This method uses a multilayer perceptron to pred…
-
New framework uses hierarchical RL for neural network compression
Researchers have developed HiReLC, a hierarchical reinforcement learning framework designed to jointly quantize and prune deep neural networks. This approach uses low-level agents for per-kernel configurations and high-…
-
Tapered Language Models improve performance by reallocating parameters
Researchers have introduced Tapered Language Models (TLMs), an architectural innovation that reallocates parameters across model layers. Instead of uniform distribution, TLMs allocate more capacity to earlier layers and…
-
Gated MLPs viewed as rank-1 approximation of bilinear attention
A new research paper proposes viewing conventional gated MLPs as a rank-1 approximation of a bilinear attention mechanism. The authors demonstrate that by moving the nonlinearity to one factor, the exchange symmetry bet…
-
New pruning method preserves LLM reasoning performance
Researchers have developed a new training-free method called Causal Attribution Pruning (CAP) to reduce the size of large language models while preserving their reasoning capabilities. CAP identifies and prunes less cri…
-
ITNet architecture unifies convolution, attention, and recurrence
Researchers have introduced ITNet, a novel neural network architecture that unifies convolution, attention, and recurrence into a single learnable integral transform. This architecture uses a learnable kernel, implement…
-
New Lie-Algebra Attention Treats Tokens as Group Elements
Researchers have introduced a novel attention mechanism called Lie-Algebra Attention, which treats tokens as elements of a matrix Lie group. This approach allows attention scores to be derived from the intrinsic geometr…
-
Pixel-Level Residual Diffusion Transformer advances 3D CT volume generation
Researchers have introduced the Pixel-Level Residual Diffusion Transformer (PRDiT), a novel framework designed for generating high-resolution 3D CT medical volumes. This model employs a two-stage approach, first using a…
-
New Temporal Operator Attention framework enhances time-series analysis
Researchers have introduced Temporal Operator Attention (TOA), a novel framework designed to improve time-series analysis by addressing limitations in standard attention mechanisms. TOA explicitly incorporates learnable…
-
New AI model predicts high-potential SMEs using public data
Researchers have developed SME-HGT, a Heterogeneous Graph Transformer framework designed to identify Small and Medium Enterprises (SMEs) with high potential for advancing in funding rounds. The model utilizes public dat…
-
Ghost Attractor Networks offer efficient sequential generation with stable latent structures
Researchers have introduced Ghost Attractor Networks (GANs), a novel dynamical decoder designed to improve sequential generation efficiency and control in large-scale models. GANs utilize a learned potential with a basi…
-
New theory links shock-wave dynamics to neural network training
Researchers have established a mathematical connection between shock-wave theory and the learning dynamics of stochastic gradient descent in artificial neural networks. By applying principles from differential geometry,…