arXiv research

New insights into the top-K sparse softmax gating function for deep learning.

Computations for the softmax function are significantly expensive when the number of output classes is large. In this paper, we present a novel softmax inference speedup method, Doubly Sparse Softmax (DS-Softmax), that leverages sparse mixture of sparse experts to efficiently retrieve top-k classes. Different from most…

In order to push the performance on realistic computer vision tasks, the number of classes in modern benchmark datasets has significantly increased in recent years. This increase in the number of classes comes along with increased ambiguity between the class labels, raising the question if top-1 error is the right perf…

Proposes differentiable and sparse top-k operators for neural networks.

Top-k error is currently a popular performance measure on large scale image classification benchmarks such as ImageNet and Places. Despite its wide acceptance, our understanding of this metric is limited as most of the previous research is focused on its special case, the top-1 error. In this work, we explore two direc…

Despite being the standard loss function to train multi-class neural networks, the log-softmax has two potential limitations. First, it involves computations that scale linearly with the number of output classes, which can restrict the size of problems we are able to tackle with current hardware. Second, it remains unc…

Introduces top-$k$ regularization for better feature selection in machine learning.

Evidential Softmax preserves multimodality in sparse probability distributions for generative models.

Neural language models have been widely used in various NLP tasks, including machine translation, next word prediction and conversational agents. However, it is challenging to deploy these models on mobile devices due to their slow prediction speed, where the bottleneck is to compute top candidates in the softmax layer…

Paper extends sparse alternatives to softmax for continuous domains, enabling efficient attention mechanisms.

Attention mechanisms have become ubiquitous in NLP. Recent architectures, notably the Transformer, learn powerful context-aware word representations through layered, multi-headed attention. The multiple heads learn diverse types of word relationships. However, with standard softmax attention, all attention heads are de…

This note considers softmax parameter estimation when little/no labeled training data is available, but a priori information about the relative geometry of class label log-odds boundaries is available. It is shown that `data-free' softmax model synthesis corresponds to solving a linear system of parameter equations, wh…

New method uses temperature to control sparse MoE convergence rates.

Low-rank framework for task-specific LLM ranking from sparse comparisons.

Variance reduction methods such as SVRG and SpiderBoost use a mixture of large and small batch gradients to reduce the variance of stochastic gradients. Compared to SGD, these methods require at least double the number of operations per update to model parameters. To reduce the computational cost of these methods, we i…

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…

In probabilistic classification, a discriminative model based on the softmax function has a potential limitation in that it assumes unimodality for each class in the feature space. The mixture model can address this issue, although it leads to an increase in the number of parameters. We propose a sparse classifier base…

DSelect-k improves MoE models for multi-task learning with better performance and smoother training.

Paper analyzes trade-offs in top-k classification accuracies and proposes a new loss function.

Class ambiguity is typical in image classification problems with a large number of classes. When classes are difficult to discriminate, it makes sense to allow k guesses and evaluate classifiers based on the top-k error instead of the standard zero-one loss. We propose top-k multiclass SVM as a direct method to optimiz…

The top-$k$ error is often employed to evaluate performance for challenging classification tasks in computer vision as it is designed to compensate for ambiguity in ground truth labels. This practical success motivates our theoretical analysis of consistent top-$k$ classification. Surprisingly, it is not rigorously und…

Converting an n-dimensional vector to a probability distribution over n objects is a commonly used component in many machine learning tasks like multiclass classification, multilabel classification, attention mechanisms etc. For this, several probability mapping functions have been proposed and employed in literature s…

The paper analyzes convergence rates of softmax gating in MoE models.

Paper introduces a new loss function for deep imbalanced classification.

Smoothed top-k operator improves model training efficiency.

Paper introduces efficient top-k selection with differential privacy.

Spark Transformer achieves high sparsity in FFN and attention without sacrificing model quality.

Motivated by applications in recommender systems, web search, social choice and crowdsourcing, we consider the problem of identifying the set of top $K$ items from noisy pairwise comparisons. In our setting, we are non-actively given $r$ pairwise comparisons between each pair of $n$ items, where each comparison has noi…

Paper characterizes minimax regret rates for online ranking with top-k feedback.

Unified model for prediction and deferral selects top-k entities efficiently.

Paper introduces MPES for top-k ranking BO with preferential observations.

We present online boosting algorithms for multilabel ranking with top-k feedback, where the learner only receives information about the top k items from the ranking it provides. We propose a novel surrogate loss function and unbiased estimator, allowing weak learners to update themselves with limited information. Using…

Study improves top-k set prediction with low cardinality.

Paper extends top-k Mallows model for better user preference analysis.

Transformers with linear space and time complexity for accurate attention estimation.

Sparse attention model reduces long-context inference time with exponential accuracy guarantees.

SAE-FiRE extracts key financial info from long documents, improving earnings surprise predictions.

This paper compares average-K and top-K classification methods under ambiguity.

Study on top-$k$ classification with new loss functions and algorithms.

This paper explores the preference-based top-$K$ rank aggregation problem. Suppose that a collection of items is repeatedly compared in pairs, and one wishes to recover a consistent ordering that emphasizes the top-$K$ ranked items, based on partially revealed preferences. We focus on the Bradley-Terry-Luce (BTL) model…

The paper proposes methods to identify and sample from mixtures of Mallows models for top-k rankings.

Recent literature on unsupervised learning focused on designing structural priors with the aim of learning meaningful features, but without considering the description length of the representations. In this thesis, first we introduce the $\varphi$ metric that evaluates unsupervised models based on their reconstruction …

Unified framework for deferring queries to top-k experts, improving accuracy-cost trade-offs.

This paper develops sparse alternatives to continuous distributions, including new types of Gaussians and attention mechanisms.

We propose the Limited Multi-Label (LML) projection layer as a new primitive operation for end-to-end learning systems. The LML layer provides a probabilistic way of modeling multi-label predictions limited to having exactly k labels. We derive efficient forward and backward passes for this layer and show how the layer…

Previous literature on unsupervised learning focused on designing structural priors with the aim of learning meaningful features. However, this was done without considering the description length of the learned representations which is a direct and unbiased measure of the model complexity. In this paper, first we intro…

Motivation: How do we integratively analyze large-scale multi-platform genomic data that are high dimensional and sparse? Furthermore, how can we incorporate prior knowledge, such as the association between genes, in the analysis systematically? Method: To solve this problem, we propose a Scalable Network Constrained T…

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…