Paper proposes graph-based separable transforms for video coding.
problem Improving video coding efficiency by better capturing residual block statistics.
method Derives graph-based separable transforms (GBSTs) from line graphs with weights determined by parameters.
result GBSTs achieve about 0.4% average coding gain over existing transforms in VVC.
Transformer model improves source code summarization.
problem Generating readable summaries of source code.
method Transformer model with self-attention mechanism for code representation.
result Transformer model outperforms state-of-the-art techniques.
Model learns code representations from comments for data analysis tasks.
problem Lack of descriptive labels for analyzing large code corpora.
method Weakly supervised transformer architecture for joint code and comment representation.
result Model achieves 38% accuracy increase over expert-supplied heuristics.
Paper presents a code authorship attribution attack using adversarial learning.
problem Misleading attribution of source code using machine learning methods.
method Exploits adversarial examples and semantics-preserving code transformations guided by Monte-Carlo tree search.
result Demonstrates substantial effect on attribution methods, reducing accuracy from over 88% to 1%.
Transformer models waste resources on long-context tasks.
problem Redundant attention computations in Transformer models for long-context tasks.
method Reformulate sequence modeling as supervised learning, analyze attention sparsity, formulate attention optimization as linear coding problem, propose Dynamic Group Attention.
result DGA reduces computational costs while maintaining performance.
Tree-Transformer improves grammar correction in code and natural language.
problem Grammar correction in tree-structured data.
method Tree-Transformer neural network architecture for tree-structured data.
result Significant improvement in grammar correction accuracy.
Probabilistic inference procedures are usually coded painstakingly from scratch, for each target model and each inference algorithm. We reduce this effort by generating inference procedures from models automatically. We make this code generation modular by decomposing inference algorithms into reusable program-to-progr…
New analysis tightens memory capacity of Hopfield models using spherical codes.
problem Optimizing memory capacity in modern Hopfield models and Kernelized Hopfield Models.
method Connecting Hopfield models to spherical codes in information theory, establishing an optimal capacity bound and a sub-linear algorithm.
result First tight and optimal asymptotic memory capacity for modern Hopfield models, matching known lower bounds.
Automatic differentiation (AD) is an essential primitive for machine learning programming systems. Tangent is a new library that performs AD using source code transformation (SCT) in Python. It takes numeric functions written in a syntactic subset of Python and NumPy as input, and generates new Python functions which c…
Generative AI decodes quantum codes without labeled data.
problem Efficient decoding of quantum error-correcting codes.
method Generative Transformers learn logical operators from unsupervised syndromes.
result Significantly better decoding accuracy than traditional methods.
Predictive Sparse Manifold Transform learns dynamic video sequences.
problem Learning and predicting natural dynamics in video sequences.
method Two-layer framework: sparse coding and manifold learning.
result PSMT with dynamic embedding space outperforms static baselines in future frame prediction.
Study makes code models robust to small changes that keep functionality.
problem Vulnerability of deep neural networks to adversarial examples in source code.
method Defined a powerful adversary and adversarial training to learn robust models.
result Significant gains in robustness demonstrated across different languages and architectures.
A new iterative low complexity algorithm has been presented for computing the Walsh-Hadamard transform (WHT) of an N dimensional signal with a K-sparse WHT, where N is a power of two and K=O(Nα), scales sub-linearly in N for some 0<α<1. Assuming a random support model for the non-zero transform domain…
COSET benchmarks neural program embeddings using diverse source-code datasets.
problem Evaluating neural program embeddings is challenging due to lack of straightforward metrics.
method COSET framework with labeled programs, transformations, and a pilot study.
result COSET identifies strengths and weaknesses of neural models and program characteristics.
ERP improves drug discovery by balancing molecule generation quality and efficiency.
problem Generating valid and optimal molecules from large language models.
method Entropy-Reinforced Planning (ERP) for Transformer Decoding.
result ERP outperforms current state-of-the-art algorithms by 1-5 percent on SARS-CoV-2 and human cancer cell targets.
The need to efficiently calculate first- and higher-order derivatives of increasingly complex models expressed in Python has stressed or exceeded the capabilities of available tools. In this work, we explore techniques from the field of automatic differentiation (AD) that can give researchers expressive power, performa…
We apply neural nets with ReLU gates in online reinforcement learning. Our goal is to train these networks in an incremental manner, without the computationally expensive experience replay. By studying how individual neural nodes behave in online training, we recognize that the global nature of ReLU gates can cause und…
DiffTaichi enables fast, differentiable physical simulations with shorter code.
problem Building efficient differentiable physical simulators.
method Differentiable programming language (DiffTaichi) that generates gradients using source code transformations and a light-weight tape.
result Differentiable physical simulators written in DiffTaichi are faster and more concise than existing methods.
We present a signal representation framework called the sparse manifold transform that combines key ideas from sparse coding, manifold learning, and slow feature analysis. It turns non-linear transformations in the primary sensory signal space into linear interpolations in a representational embedding space while maint…
Paper develops multilingual job classification for ISCO and KZiS.
problem Classifying job advertisements for accurate occupation coding.
method Transformer architecture for hierarchical multi-class classification.
result Hierarchical structure improves prediction accuracy by 1-2 percentage points.
Paper models and compresses wideband CSI feedback in FDD MIMO systems.
problem Fundamental limits of channel state information (CSI) feedback in FDD massive MIMO systems.
method Modeling CSI as a Gaussian-mixture source with latent geometry states, proposing Gaussian-mixture transform coding (GMTC).
result Near-optimal CSI compression achieved through state-adaptive transform coding without large neural encoders.
This paper clarifies VAE's property through geometric and information-theoretic interpretations.
problem The transparency of VAE model is an underlying issue.
method Quantitative understanding of VAE through differential geometry and information theory.
result VAE can be mapped to an implicit isometric embedding with a scale factor derived from the posterior parameter.
The wavelet scattering transform is an invariant signal representation suitable for many signal processing and machine learning applications. We present the Kymatio software package, an easy-to-use, high-performance Python implementation of the scattering transform in 1D, 2D, and 3D that is compatible with modern deep …
B-cos transformers explain Vision Transformers' decisions.
problem Lack of holistic explanations for transformer outputs.
method Formulate each component as dynamic linear, allowing a single linear transform for summarization.
result Bcos-ViTs are highly interpretable and competitive on ImageNet.
Deep learning aids ADMM-based decoding for binary linear codes.
problem Improving decoding efficiency for binary linear codes.
method Designing a decoding network based on ADMM and deep learning.
result Numerical results show improved performance compared to original ADMM.
Paper trains a Transformer to add numbers of any length.
problem Training Transformers to handle arbitrary-length addition.
method Autoregressive generation from right to left.
result Trains a Transformer to generalize addition of numbers of any length.
I have submitted a new version to arXiv:1910.13806. I forget to choose to replace the old version, but submitted a new one. It's my mistake.
UGformer uses transformers to learn graph representations.
problem Graph representation learning for various tasks.
method UGformer is a transformer-based GNN model that samples or considers all neighbors for each node.
result UGformer achieves state-of-the-art accuracy on graph classification and text classification tasks.
TAPER learns unified patient EHR representations for healthcare tasks.
problem Irregular and multimodal data in electronic health records.
method Transformer networks and BERT for embedding structured and unstructured data.
result TAPER model outperforms on mortality, readmission, and length of stay tasks.
Minimalistic unsupervised learning with sparse manifold transform achieves SOTA performance.
problem Achieving state-of-the-art unsupervised learning performance without complex engineering.
method Sparse manifold transform, leveraging sparse coding, manifold learning, and slow feature analysis.
result 99.3% KNN top-1 accuracy on MNIST, 81.1% on CIFAR-10, and 53.2% on CIFAR-100.
Jukebox generates high-fidelity songs with singing in raw audio.
problem Generating music with singing in raw audio.
method Multi-scale VQ-VAE for compression, autoregressive Transformers for modeling.
result Generates high-fidelity and diverse songs with coherence up to multiple minutes.
Recent progress on deep learning has made it possible to automatically transform the screenshot of Graphic User Interface (GUI) into code by using the encoder-decoder framework. While the commonly adopted image encoder (e.g., CNN network), might be capable of extracting image features to the desired level, interpreting…
BayesDLL offers a PyTorch library for Bayesian deep learning with large models.
problem Bayesian inference for large-scale deep networks.
method Variational inference, MC-dropout, stochastic-gradient MCMC, Laplace approximation.
result BayesDLL can handle Vision Transformers and pre-trained model weights as priors.
Most neural-network based speaker-adaptive acoustic models for speech synthesis can be categorized into either layer-based or input-code approaches. Although both approaches have their own pros and cons, most existing works on speaker adaptation focus on improving one or the other. In this paper, after we first systema…
ETC improves Transformer models for long and structured inputs.
problem Scaling input length and encoding structured inputs in Transformers.
method Introduces global-local attention, relative position encodings, and CPC pre-training.
result Achieves state-of-the-art results on four natural language datasets.
DiffC compresses images by diffusing Gaussian noise, outperforming state-of-the-art methods.
problem Efficient lossy image compression without transform coding.
method Unconditional diffusion generative models to encode and denoise corrupted pixels.
result DiffC outperforms HiFiC on ImageNet 64x64, achieving a 3 dB gain at high bitrates.
A new feature coding method for invariant features using tensor products.
problem Learning invariant features for transformations represented by orthogonal matrices.
method Group-invariant feature vector using tensor-product representations of basic representations.
result Group-invariant feature vector contains sufficient discriminative information for linear classifiers.
This work shows how transformers use multi-concept word semantics for efficient in-context learning.
problem Understanding the connection between transformer-based LLMs' multi-concept semantic representation and their innovative in-context learning abilities.
method A concept-based low-noise sparse coding prompt model, leveraging advanced techniques to analyze the exponential convergence of 0-1 loss over non-convex training dynamics.
result Transformers leverage multi-concept word semantics to enable powerful and excellent out-of-distribution in-context learning.
A new deep learning approach enhances signature transforms for data analysis.
problem Improving data analysis through more flexible signature transforms.
method Combining deep learning with signature transforms to select terms data-dependently.
result Signature transform can be used as a pooling operation within neural networks.
Paper introduces graph-based transforms for video compression.
problem Efficiently represent video signals for compression.
method Develops two techniques for designing graph-based transforms (GL-GBTs and EA-GBTs).
result Graph-based transforms outperform traditional KLT in video compression.
Sparse coding is a common approach to learning local features for object recognition. Recently, there has been an increasing interest in learning features from spatio-temporal, binocular, or other multi-observation data, where the goal is to encode the relationship between images rather than the content of a single ima…
Enhanced Transformer solves math problems better with explicit relation encoding.
problem Improving Transformer models for solving math word problems.
method Integrates Tensor-Product Representations and TP-Attention mechanism.
result Sets new state of the art on the Mathematics Dataset.
We introduce a differential geometric framework for describing families of quantum error-correcting codes and for understanding quantum fault tolerance. This work unifies the notion of topological fault tolerance with fault tolerance in other kinds of quantum error-correcting codes. In particular, we use fibre bundles …
Improves demand forecasting accuracy through recurrent transform learning.
problem Challenging task of building demand forecasting.
method Developed two versions of recurrent transform learning (RTL and R2TL) for feature extraction and regression.
result Both RTL and R2TL techniques outperform state-of-the-art methods.
GradPower speeds up language model training with minimal code changes.
problem Slower training of large language models.
method Elementwise sign-power transformation applied to gradients.
result Consistently lower terminal loss across various models and datasets.
B-cos transforms improve neural network interpretability by aligning weights.
problem Improving interpretability of deep neural networks.
method Replacing linear transforms with B-cos transforms that promote weight-input alignment during training.
result B-cos transforms lead to highly interpretable and task-relevant linear summaries of neural network computations.
Sparse codes improve optimal control tasks with correlated inputs.
problem Optimal control tasks with correlated feature inputs.
method Used a sparse code to represent natural images in an optimal control task solved with neuro-dynamic programming.
result An over-complete sparse code increases memory capacity and learning speed beyond a complete code.
Transformers struggle to approximate smooth functions, relying on piecewise constant approximations.
problem Understanding the expressivity of Transformers for function approximation.
method Theoretical analysis and experimental validation of Transformer's ability to approximate smooth functions.
result Transformers cannot reliably approximate smooth functions, relying on piecewise constant approximations.