We propose sparsemax, a new activation function similar to the traditional softmax, but able to output sparse probabilities. After deriving its properties, we show how its Jacobian can be efficiently computed, enabling its use in a network trained with backpropagation. Then, we propose a new smooth and convex loss func…
On-device research index
arXiv research
A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.
169,051 papers · 148 categories
Trend · papers per month
12 results for “sparsemax”
Paper proposes models to identify topic sparsity in social media texts.
problem Topic sparsity in online social media content.
method Sparsemax and relaxed Wasserstein for topic sparsity.
result Proposed models achieve better topic sparsity identification and performance.
Introduces Fitzpatrick losses, tighter than Fenchel-Young losses.
problem Improving loss functions for machine learning.
method Introduces Fitzpatrick losses based on the Fitzpatrick function.
result Fitzpatrick losses are tighter than Fenchel-Young losses.
GSANet improves semantic segmentation accuracy with selective and global attention.
problem Semantic segmentation accuracy improvement.
method Global and selective attention mechanism with ASPP and sparsemax.
result GSANet achieves state-of-the-art accuracy on ADE20k and Cityscapes datasets.
In this paper, we propose a novel maximum causal Tsallis entropy (MCTE) framework for imitation learning which can efficiently learn a sparse multi-modal policy distribution from demonstrations. We provide the full mathematical analysis of the proposed framework. First, the optimal solution of an MCTE problem is shown …
Modern neural networks are often augmented with an attention mechanism, which tells the network where to focus within the input. We propose in this paper a new framework for sparse and structured attention, building upon a smoothed max operator. We show that the gradient of this operator defines a mapping from real val…
Unified framework for sparse alternatives to softmax with control over sparsity.
problem Lack of understanding and explicit control over sparsity in probability mapping functions.
method Unified framework encompassing softmax, sum-normalization, spherical softmax, and sparsemax. Two novel sparse formulations (sparsegen-lin and sparsehourglass) and convex loss functions developed.
result Improved performance in multilabel classification and seq2seq tasks like neural machine translation and abstractive summarization.
This paper develops sparse alternatives to continuous distributions, including new types of Gaussians and attention mechanisms.
problem Creating flexible continuous distributions with varying support for machine learning applications.
method Defining -regularized prediction maps and Fenchel-Young losses for arbitrary domains, and deriving new types of Gaussians and attention mechanisms.
result Sparse alternatives to continuous distributions, including deformed exponential families and -Gaussians, are introduced.
Paper extends sparse alternatives to softmax for continuous domains, enabling efficient attention mechanisms.
problem Efficiently assigning zero probability to irrelevant categories in continuous domains.
method Extend alpha-entmax to continuous domains, introducing continuous-domain attention mechanisms.
result Continuous attention allows attending to time intervals and compact regions, improving text classification, machine translation, and visual question answering.
New method for efficient marginalization of discrete latent variables in neural networks.
problem Computational challenges in training models with discrete latent variables.
method Parameterizing discrete distributions using sparse mappings (sparsemax and structured variants) to reduce support and enable efficient marginalization.
result Achieved good performance in various tasks with efficient and practical training.
Paper proposes a new method to learn EBMs and their partition function.
problem Intractability of exact MLE for EBMs due to partition function computation.
method Jointly learns an energy model and its log-partition function using neural networks.
result First tractable method for optimizing sparsemax loss in large spaces.
Subject-specific Deep Neural Networks for Count Data with High-cardinality Categorical Featurescs.LG
Proposes a new DNN framework for count data with high-cardinality features.
problem Real-world data often have correlations and high-cardinality categorical features that traditional DNNs overlook.
method Introduces a hierarchical likelihood learning framework with gamma random effects for Poisson DNNs.
result Improves prediction performance by capturing nonlinear effects and subject-specific cluster effects.