Proposes sigsoftmax to overcome the softmax bottleneck in language models.
problem Softmax function acts as a bottleneck in neural network representational capacity.
method Identifies the cause of softmax bottleneck and proposes sigsoftmax as a new activation function.
result Sigsoftmax outperforms softmax in language modeling tasks.
Proposes learnable monotonic functions to improve softmax's limitations.
problem Softmax's limited representational capacity in large output vocabularies.
method Learn parametric monotonic functions on logits.
result Improves quality metrics over traditional Linear-Softmax in language models.
LLMs learn peaked distributions slowly due to power-law losses.
problem Slow convergence of loss in training large language models.
method Systematic analysis of toy models and empirical evaluation of LLMs.
result Power-law time scaling with an exponent of 1/3 for learning peaked distributions.
Paper speeds up neural language model inference by 20x for top-k word prediction.
problem Slow inference speed of neural language models on mobile devices.
method Introduced a screening model using Gumbel softmax to approximate softmax layer.
result Achieved 20.4x speedup with 98.9% precision@1 and 99.3% precision@5 for German to English translation.
Develops RF-softmax for faster training with softmax cross entropy.
problem High computational cost of training with softmax cross entropy.
method Random Fourier Features for efficient sampling from approximate softmax distribution.
result RF-softmax provides low bias in estimating both softmax distribution and its gradient.
Hierarchical Softmax approximates class probabilities for large datasets efficiently.
problem Computational inefficiency of Softmax for large-scale classification tasks.
method Used Hierarchical Softmax to approximate class probabilities efficiently.
result Hierarchical Softmax performance degrades as the number of classes increases.
Revises logistic-softmax likelihood for Bayesian meta-learning in few-shot classification.
problem Inherent uncertainty in logistic-softmax leads to suboptimal performance in meta-learning.
method Redesigns logistic-softmax likelihood with a temperature parameter for better control of prior confidence.
result Achieves well-calibrated uncertainty estimates and comparable/superior performance on benchmark datasets.
Paper introduces Balanced Meta-Softmax for better long-tailed visual recognition.
problem Long-tailed distribution mismatch between training and testing data.
method Balanced Meta-Softmax, an unbiased extension of Softmax, using a Meta Sampler.
result Balanced Meta-Softmax outperforms state-of-the-art solutions on visual recognition and instance segmentation.
Paper introduces hierarchical softmax for global hierarchical classification tasks.
problem Improving classification accuracy in tasks with class hierarchies.
method Global hierarchical neural networks using hierarchical softmax.
result Hierarchical softmax outperforms regular softmax in multiple datasets.
DS-Softmax speeds up softmax inference by learning sparse experts.
problem Expensive softmax computations for large output classes.
method Sparse mixture of sparse experts for efficient top-k class retrieval.
result Significant computation reductions achieved at no performance loss.
Softmax emerges naturally in neural networks as a measure of conditional mutual information.
problem The artificial nature of softmax in neural networks.
method Information-theoretic perspective to derive log-softmax and evaluate conditional mutual information.
result Training deterministic neural networks through log-softmax maximises conditional mutual information.
Binary testing for softmax models requires many samples, similar to leverage score models.
problem Binary hypothesis testing for softmax models and leverage score models.
method Analyzing sample complexity and drawing analogies between models.
result Sample complexity is asymptotically \(O(ε^{-2})\), where \(ε\) is the distance between model parameters.
New algorithms make softmax optimization unbiased and scalable.
problem Efficiently computing softmax distributions with large categories.
method Proposed unbiased algorithms for maximizing softmax likelihood.
result Comprehensive outperformance on seven real-world datasets.
The paper investigates polynomial alternatives to softmax in transformer models.
problem The effectiveness of softmax attention in transformers is questioned.
method The authors explore polynomial activations as alternatives to softmax, focusing on their ability to regularize the attention matrix.
result Certain polynomials can serve as effective substitutes for softmax in transformer applications, achieving strong performance.
Softmax temperature influences model representation rank and performance.
problem Understanding and optimizing softmax function's impact on model representations.
method Investigated softmax function's role in deep neural networks, introduced rank deficit bias.
result Softmax temperature affects model representation rank and can improve performance.
Improved classifier accuracy by using more of the class-specific structure in trained models.
problem Softmax ignores valuable information encoded in the full array of class response distributions.
method Developed a hybrid classifier (Softmax-Pooling Hybrid, SPH) that uses Softmax on high-scoring samples and a log-likelihood method on low-scoring samples. result Reduces test set error by 6% to 23% using the exact same trained model.
Cross-entropy loss together with softmax is arguably one of the most common used supervision components in convolutional neural networks (CNNs). Despite its simplicity, popularity and excellent performance, the component does not explicitly encourage discriminative learning of features. In this paper, we propose a gene…
Paper proposes a new softmax loss for better performance in Positive and Unlabeled data tasks.
problem Current softmax losses and sampling schemes have drawbacks in Positive and Unlabeled learning.
method Proposes Relaxed Softmax (RS) loss and a new negative sampling scheme.
result New training objective drives uplifts in performance on textual and recommendation datasets.
Softmax confidence misrepresents uncertainty in neural networks.
problem Neural networks fail to increase uncertainty on out-of-distribution data.
method Investigates two implicit biases in softmax confidence.
result Softmax confidence correlates with epistemic uncertainty due to decision boundary structure and deep network filtering.
Unified framework for studying softmax attention under large prompts.
problem Challenges in theoretical analysis of softmax attention.
method Measure-based framework for finite and infinite prompts.
result Softmax attention converges to linear attention in the large-prompt regime.
In a multi-class classification problem, it is standard to model the output of a neural network as a categorical distribution conditioned on the inputs. The output must therefore be positive and sum to one, which is traditionally enforced by a softmax. This probabilistic mapping allows to use the maximum likelihood pri…
Estimates softmax parameters without data, using class geometry.
problem Softmax parameter estimation with limited labeled data.
method Solves linear equations based on class geometry specifications.
result Closed-form solutions possible without data sampling.
Transformers use ReLUs to approximate softmax efficiently.
problem Analyzing resource usage in softmax transformer models.
method Translating ReLU approximation results to softmax attention mechanisms.
result Economic resource bounds for softmax attention mechanisms.
Paper shows softmax output misleads in evaluating adversarial example strength.
problem Softmax output misleads in evaluating adversarial example strength.
method Demonstrates how adversarial examples can exploit softmax properties.
result Softmax output is a poor indicator of adversarial example strength.
Improved SincNet for better speaker recognition.
problem Speaker recognition challenges and the need for better deep learning models.
method Proposes AM-SincNet, a SincNet-based model with an improved AM-Softmax layer.
result Improved speaker recognition performance, achieving a 40% Frame Error Rate reduction.
New method adds uncertainty estimation to softmax outputs.
problem Uncertainty estimation in neural networks.
method Extend softmax layer with an additional constant input.
result Performs comparably to more computationally expensive methods.
Solves challenges in estimating parameters of softmax gating Gaussian mixture models.
problem Identifiability issues and complex interactions in Gaussian mixture of experts.
method Proposes novel Voronoi loss functions and establishes convergence rates of MLE.
result Connects convergence rate of MLE to a solvability problem of polynomial equations.
Chronicals boosts LLM fine-tuning speed by 3.51x over Unsloth.
problem Memory bottleneck in fine-tuning large language models.
method Four optimizations: fused kernels, cross-entropy reduction, LoRA+, and sequence packing.
result 3.51x speedup on A100-40GB with Qwen2.5-0.5B.
DBS improves convergence in reinforcement learning.
problem Softmax operator convergence issues in reinforcement learning.
method Dynamic Boltzmann Softmax (DBS) updates value function.
result DBS enables better value function estimation and convergence.
A softmax operator applied to a set of values acts somewhat like the maximization function and somewhat like an average. In sequential decision making, softmax is often used in settings where it is necessary to maximize utility but also to hedge against problems that arise from putting all of one's weight behind a sing…
The paper proves neural networks with ReLU and softmax can approximate any function.
problem Approximating functions and class labels in neural networks.
method Extended universal approximator theory to neural networks with ReLU and softmax.
result Neural networks with ReLU and softmax can approximate any function and class labels.
LUNA improves linear attention for long sequences without sacrificing accuracy.
problem Quadratic computational cost of softmax attention in long-sequence domains.
method LUNA learns a learnable kernel feature map to reduce attention cost to linear while maintaining accuracy.
result LUNA achieves state-of-the-art performance on the LRA and excels at post-hoc conversion.
Introduces gradient decay in Softmax for better generalization.
problem Improving generalization performance in neural networks.
method Gradient decay hyperparameter in Softmax for varying gradient rates based on probability.
result Gradient decay rate affects generalization performance and can be tuned for better optimization.
Unified interpretation of softmax cross-entropy and negative sampling for knowledge graph embedding.
problem Lack of theoretical relationship between softmax cross-entropy and negative sampling loss functions in knowledge graph embedding.
method Used Bregman divergence to provide a unified interpretation of the two loss functions.
result Theoretical findings for fair comparison of softmax cross-entropy and negative sampling are derived.
New L2 regularization improves softmax MAB performance.
problem Improving softmax MAB performance with vanishing regularization.
method L2 regularization with vanishing parameter analyzed and proven convergent.
result Vanishing L2 regularization makes softmax MAB more numerically advantageous.
Gradient flow in softmax models tends to produce low-entropy outputs.
problem Understanding the training dynamics of softmax-based models.
method Analysis of gradient flow dynamics in the value-softmax model.
result Gradient flow drives optimization towards low-entropy solutions.
Softmax PG methods can take extremely long to converge, even with exact gradients.
problem Softmax policy gradient methods can take an impractically long time to converge in reinforcement learning.
method Softmax policy gradient methods with exact gradient computation.
result Softmax PG methods can take exponential number of iterations to converge, even with optimal initialization.
The paper identifies universal features for high-dimensional data inference.
problem Identifying universal low-dimensional features from high-dimensional data for inference tasks.
method Introduces natural notions of universality and shows a local equivalence among them, using information geometry.
result Reveals the complementary roles of various data analysis techniques.
Softmax Bellman operator improves Q-function performance in RL despite sub-optimality.
problem Softmax Bellman operator's impact on value functions in RL is problematic.
method Revisited theoretical properties of softmax Bellman operator, proving convergence and overestimation reduction.
result Softmax Bellman operator leads to superior policies in practice, even outperforming double Q-learning.
Density-Softmax improves uncertainty estimation and robustness without sampling, reducing model size and latency.
problem Sampling-based uncertainty estimation methods suffer from large model size and high latency.
method Combines a Lipschitz-constrained feature extractor with the softmax layer to create a sampling-free deterministic framework.
result Density-Softmax reduces over-confidence under distribution shifts and achieves competitive results in uncertainty and robustness.
WEINCE improves contrastive learning by correcting softmax biases.
problem Softmax in InfoNCE can lead to misaligned statistical assumptions in contrastive learning.
method WEINCE uses anchor-wise online batch statistics to blend softmax logits with an endpoint shortfall correction.
result WEINCE yields consistent improvements in frozen-feature evaluation across five vision benchmarks.
Generative Adversarial Networks (GAN) have limitations when the goal is to generate sequences of discrete elements. The reason for this is that samples from a distribution on discrete objects such as the multinomial are not differentiable with respect to the distribution parameters. This problem can be avoided by using…
Softmax is found ineffective for NL block, leading to improved performance.
problem Inefficiency of softmax in NL block for global context modeling.
method Empirical analysis and replacement of softmax with scaling factor.
result Improved performance on various datasets with reduced computational cost.
ZeroS improves Transformers by adding negative weights, matching or beating softmax attention.
problem Limited performance of linear attention methods, especially in long context sequences.
method Proposes Zero-Sum Linear Attention (ZeroS) that removes the zero-order term and reweights zero-sum softmax residuals.
result ZeroS matches or exceeds standard softmax attention across various benchmarks, theoretically expanding representable functions.
New insights into the top-K sparse softmax gating function for deep learning.
problem Understanding the theoretical effects of the top-K sparse softmax gating function on density and parameter estimations.
method Using a Gaussian mixture of experts, novel loss functions, and theoretical analysis.
result The convergence rates of density and parameter estimations are parametric under certain conditions, but slow under over-specified models.
Proposes a new neural network architecture inspired by biology to improve learning and information flow.
problem Improving artificial neural networks to match biological neuron properties like multidirectional propagation and probabilistic modeling.
method Extends KAN approach with joint distribution neurons that can propagate values and distributions, including variance and higher-order moments.
result Proposed architecture can predict and propagate distributions, including expected values and variances.
A new method speeds up SoftMax normalization for embedding learning.
problem Efficiently learning distributed representations with SoftMax normalization.
method Proposes a linear-time heuristic approximation for mSoftMax(XYT), optimizing cross entropy. result Achieves higher or comparable accuracy to existing methods with lower computational time.
Heating up softmax improves feature compactness for better metric learning.
problem Learning embeddings for samples of the same category to be compact while different categories spread out.
method Training classifiers with different temperature values of softmax function, then heating up the classifier.
result Classifier with increasing temperatures achieves state-of-the-art performance on metric learning benchmarks.