Improved graph attention model for noisy graphs.
problem Understanding and improving graph attention in noisy graphs.
method Proposes SuperGAT, a self-supervised graph attention network.
result SuperGAT learns more expressive attention by encoding edges.
Paper introduces a conformer-based system for streaming language identification in long-form speech.
problem Language identification in long-form audio.
method Conformer layers with attentive temporal pooling and domain adaptation.
result Conformer-based models significantly outperform LSTM and transformer models.
Express improves causal attention guarantees for language models.
problem Improving causal attention guarantees for language models.
method Introducing Express, a tool for converting non-causal attention into causal with matching guarantees.
result Express improves causal attention guarantees to log^(3/2)(n)/s with minimal memory and compression overhead.
We introduce the 2-simplicial Transformer, an extension of the Transformer which includes a form of higher-dimensional attention generalising the dot-product attention, and uses this attention to update entity representations with tensor products of value vectors. We show that this architecture is a useful inductive …
New insights into how encoder-decoder networks generate attention matrices.
problem Understanding how encoder-decoder networks use attention matrices.
method Decomposing hidden states into temporal and input-driven components.
result Attention matrices are formed based on task requirements, not architecture type.
A framework for transformer attention layers derived from SVR.
problem Developing principled attention mechanisms for transformers.
method Mapping self-attention to SVR, deriving new attention types.
result Improved transformer performance and efficiency.
Bayesian theory explains abrupt emergence of copy subcircuit in attention.
problem Understanding the abrupt emergence of the copy subcircuit in attention during training.
method Deriving a closed-form posterior over the attention matrix and reducing it to a low-dimensional order parameter space.
result Derive a phase transition in the amount of training data.
In this paper, we present a novel approach for incorporating external knowledge in Recurrent Neural Networks (RNNs). We propose the integration of lexicon features into the self-attention mechanism of RNN-based architectures. This form of conditioning on the attention distribution, enforces the contribution of the most…
ABIForest improves anomaly detection using attention weights.
problem Anomaly detection in datasets.
method Attention mechanism integrated into Isolation Forest.
result ABIForest outperforms standard Isolation Forest on synthetic and real datasets.
Self attention mechanisms have become a key building block in many state-of-the-art language understanding models. In this paper, we show that the self attention operator can be formulated in terms of 1x1 convolution operations. Following this observation, we propose several novel operators: First, we introduce a 2D ve…
AIM models explain deep learning in attention layers, offering solvable insights.
problem Understanding how deep learning models learn in attention layers.
method Statistical mechanics and random matrix theory.
result Closed-form predictions for Bayes-optimal generalization error and gradient descent performance.
In (Yang et al. 2016), a hierarchical attention network (HAN) is created for document classification. The attention layer can be used to visualize text influential in classifying the document, thereby explaining the model's prediction. We successfully applied HAN to a sequential analysis task in the form of real-time m…
Study proposes a statistical test for Vision Transformer's attention mechanisms.
problem ViT's attention mechanisms may focus on irrelevant regions, leading to unreliable evidence.
method Selective inference framework to quantify statistical significance of attentions as p-values.
result Proposed method enables reliable quantification of false positive detection probability of attentions.
We incorporate Tensor-Product Representations within the Transformer in order to better support the explicit representation of relation structure. Our Tensor-Product Transformer (TP-Transformer) sets a new state of the art on the recently-introduced Mathematics Dataset containing 56 categories of free-form math word-pr…
Streamable model improves speech recognition performance.
problem Improving speech recognition in long-form audio.
method Chunk-based attention model for streaming speech recognition.
result Streamable model maintains competitive performance and generalizes well to long-form speech.
Existing attention mechanisms are trained to attend to individual items in a collection (the memory) with a predefined, fixed granularity, e.g., a word token or an image grid. We propose area attention: a way to attend to areas in the memory, where each area contains a group of items that are structurally adjacent, e.g…
A new perspective on self-attention models using MLPs.
problem Improving sequence modeling with self-attention mechanisms.
method Introducing HyperMLP and HyperGLU, which use dynamic two-layer MLPs with reverse-offset layout.
result HyperMLP/HyperGLU consistently outperform softmax-attention baselines.
ABRF uses attention weights to improve RF performance.
problem Improving the performance of random forest models.
method Attention mechanism applied to random forest with optimization and gradient-based methods.
result The proposed ABRF models outperform standard RF models on various datasets.
SDPA is shown to be an optimal transport problem in deep learning.
problem The mathematical foundation and optimization perspective of SDPA.
method SDPA is shown to be the exact solution to a degenerate, one-sided Entropic Optimal Transport (EOT) problem.
result The SDPA mechanism is a principled mechanism where the forward pass performs optimal inference and the backward pass implements a rational, manifold-aware learning update.
Transformer++ improves neural machine translation BLEU scores.
problem Handling long-range dependencies in sentences.
method Proposes a new multi-head attention mechanism with context learning.
result Achieves new state-of-the-art BLEU scores on English-to-German and English-to-French translation tasks.
OSA overcomes instability in skipless Transformers.
problem Instability in skipless Transformers using Softmax Self-Attention.
method OSA parametrizes attention matrix to be orthogonal via skew-symmetric matrix exponential.
result OSA allows for training non-causal Transformers without skip connections and normalisation layers.
Efficiently accelerates attention calculation for Transformers with relative positional encoding.
problem Quadratic complexity of attention in long sequences.
method Kernelized attention with Fast Fourier Transform (FFT) for RPE.
result Achieves O(n log n) time complexity, mitigates training instability, and outperforms other models.
Attention temperature improves robustness of ICL in high-dimensional settings.
problem ICL robustness failure under distribution shift in high dimensions.
method Analyzed a Transformer with approximate softmax attention, derived a closed-form error expression, and showed optimal temperature minimizes error.
result Optimal attention temperature minimizes ICL generalization error under distribution shift.
Attention tokens are group elements, negated algebra norms.
problem Attention mechanism for matrix Lie groups
method Lie-Algebra Attention
result Score matches learned kernel on invariant poses, outperforms vector-token baseline
OLS predictions are shown to be similar to attention mechanisms in models.
problem OLS in traditional statistics and econometrics.
method Rewriting OLS as an attention mechanism in a transformed space.
result OLS can be understood as minimizing squared prediction errors via optimal embedding and decoding.
Acoustic scenes are rich and redundant in their content. In this work, we present a spatio-temporal attention pooling layer coupled with a convolutional recurrent neural network to learn from patterns that are discriminative while suppressing those that are irrelevant for acoustic scene classification. The convolutiona…
Unified framework for non-linear attention using modern Hopfield networks.
problem Improving transformer model's understanding of complex relationships and efficiency.
method Proposes an energy functional based on Modern Hopfield Networks (MNH) to unify linear and non-linear attention mechanisms.
result Context wells encapsulate contextual relationships among tokens, offering a richer representation of non-linear data.
HKT improves sequence processing with multi-scale attention and kernel analysis.
problem Processing sequences at multiple scales with efficient attention mechanisms.
method Trainable causal downsampling and convex weights for level-specific score matrices.
result HKT achieves consistent gains over standard attention across various tasks.
Transformers can emulate various algorithms by prompting, proving universality.
problem How to emulate algorithms using fixed-weight Transformers.
method Two modes of in-context algorithm emulation: task-specific and prompt-programmable. Constructing prompts that encode algorithm parameters into token representations.
result Fixed-weight Transformers can emulate a broad class of algorithms via prompts.
New method prevents entropy collapse in Transformer training, leading to more stable and robust models.
problem Training instability in Transformers, especially in attention layers.
method Spectral normalization with a learned scalar to prevent entropy collapse.
result Prevents entropy collapse, leading to more stable training.
Paper analyzes infinite-width attention layers using Tensor Programs.
problem Capturing the infinite-width limit of attention layers.
method Tensor Programs framework to rigorously identify the limit distribution.
result Derives exact form of infinite-width limit distribution without Gaussian approximations.
We demystify attention patterns in multi-head softmax models for linear data.
problem Understanding the training dynamics and emergent patterns in multi-head softmax attention models.
method Extensive empirical experiments and rigorous theoretical analysis.
result Multi-head softmax attention models approximate a debiased gradient descent predictor, outperforming single-head attention and achieving near-Bayesian optimality.
The study compares feed-forward and attention layers in language models.
problem Understanding the role of feed-forward and attention layers in language models.
method Empirical and theoretical analysis in a synthetic setting.
result Feed-forward layers learn simple distributional associations, while attention layers focus on in-context reasoning.
Develops a framework for identifying mispriced assets through attention factors for statistical arbitrage.
problem Identifying mispriced assets in statistical arbitrage trading.
method Uses conditional latent factors learned from firm characteristic embeddings to identify time-series signals and form a trading strategy.
result Achieves an out-of-sample Sharpe ratio above 4 on the largest U.S. equities over a 24-year period.
Transformers mimic Bayesian reasoning in controlled settings, revealing geometric mechanisms.
problem Verifying if transformers perform Bayesian reasoning rigorously in natural data.
method Constructing Bayesian wind tunnels with known posteriors and proving memorization impossibility.
result Transformers achieve 10−3-10−4 bit accuracy in Bayesian posteriors, while MLPs fail. Traditional sequence-to-sequence (seq2seq) models and other variations of the attention-mechanism such as hierarchical attention have been applied to the text summarization problem. Though there is a hierarchy in the way humans use language by forming paragraphs from sentences and sentences from words, hierarchical mod…
Sequential modelling with self-attention has achieved cutting edge performances in natural language processing. With advantages in model flexibility, computation complexity and interpretability, self-attention is gradually becoming a key component in event sequence models. However, like most other sequence models, self…
Transformers show strengths and weaknesses in complexity analysis.
problem Understanding the strengths and limitations of attention layers in transformers.
method Analysis of representation power through complexity parameters and task-specific constructions.
result Transformers can solve sparse averaging tasks with logarithmic complexity, but triple detection tasks require linear complexity.
ConViT combines CNN and ViT strengths, improving image classification.
problem Combining the strengths of CNNs and ViTs while avoiding their limitations.
method Introducing GPSA, a form of positional self-attention with a soft convolutional inductive bias.
result ConViT outperforms DeiT on ImageNet while offering improved sample efficiency.
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.
Crypto markets show negative spillovers between chains, not positive co-movements.
problem Negative spillovers in crypto asset returns across different blockchains.
method On-chain data from multiple blockchains (Ethereum, Solana, Binance, Arbitrum, Avalanche) analyzed over 2022-2025.
result Surges on one chain often coincide with declines on others, especially during attention shocks.
Dynamic portfolio strategy using generative model with attention mechanism.
problem Dynamic modeling of multivariate stock returns with tail-side properties.
method Dynamic generative factor model using Attention-GRU network for dynamic learning and forecasting.
result The proposed model leads to wiser investments with higher reward-risk ratios and lower tail risks.
Graph convolutional network (GCN) is generalization of convolutional neural network (CNN) to work with arbitrarily structured graphs. A binary adjacency matrix is commonly used in training a GCN. Recently, the attention mechanism allows the network to learn a dynamic and adaptive aggregation of the neighborhood. We pro…
KATA improves associative recall by optimizing feature maps derived from nonnegative attention weights.
problem Linear attention's poor performance on associative recall tasks.
method Formulates attention recall as a spherical-packing problem and introduces Kernelized Linear Attention Activations (KATA).
result KATA features offer a favorable capacity-interference tradeoff, enabling efficient associative recall.
Study SGD dynamics in sequence models, revealing training phases and influence of sequence length.
problem Understanding SGD in sequence models like attention networks.
method Derived closed-form population loss and analyzed SGD dynamics for SSI models.
result Two distinct training phases: escape from uninformative initialization and alignment with target subspace.
Language models fail to process hallucinated responses, and this study diagnoses the failure.
problem Language models fail to process hallucinated responses, leading to over-concentration or diffuse attention.
method The study uses forced scoring of benchmark-labeled responses to compute attention shapes and analyze the symmetric component of the degree-normalized attention operator.
result The study proves that every transpose-invariant spectral diagnostic of the attention operator is orientation-blind and bounds the sensitivity of any Lipschitz diagnostic by the asymmetry coefficient \(G\).
New analysis shows how cross-entropy training shapes attention in transformers.
problem Understanding how gradient-based learning creates the required internal geometry in transformers.
method Developed a first-order analysis of cross-entropy training effects on attention scores and values in a transformer attention head.
result Introduced an advantage-based routing law and responsibility-weighted update for attention scores and values, respectively.
Memory networks are neural networks with an explicit memory component that can be both read and written to by the network. The memory is often addressed in a soft way using a softmax function, making end-to-end training with backpropagation possible. However, this is not computationally scalable for applications which …