DIMCO learns discrete codes maximizing mutual info with labels, reducing overfitting and improving efficiency.
problem Learning compact discrete representations for data.
method DIMCO learns k-way d-dimensional codes maximizing mutual information between codes and labels, with regularization to enforce code independence.
result DIMCO reduces overfitting and improves efficiency in few-shot classification.
Proposes a non-parametric method for deep discrete latent variable models.
problem Learning sparse discrete latent representations in deep models.
method Iterative algorithm with Beta-Bernoulli process prior and local data scaling.
result Improves sparsity and scalability of deep discrete latent variable models.
Discrete-AIR model identifies objects in images with interpretable latent codes.
problem Identifying objects in images without labeled data.
method Recurrent Auto-Encoder with structured latent distributions for discrete, continuous, and spatial attention.
result Discrete-AIR model uses minimal latent variables for efficient inference.
Efficient Discrete Supervised Hashing improves cross-modal retrieval by preserving semantic correlations and reducing quantization error.
problem Challenges in preserving semantic correlations and reducing quantization error in cross-modal hashing for large-scale retrieval.
method Collective matrix factorization on heterogenous features and semantic embedding with class labels to learn hash codes efficiently.
result EDSH produces superior performance in both accuracy and scalability over existing methods.
Embedding methods such as word embedding have become pillars for many applications containing discrete structures. Conventional embedding methods directly associate each symbol with a continuous embedding vector, which is equivalent to applying linear transformation based on "one-hot" encoding of the discrete symbols. …
DD-VAE uses deterministic decoding for better latent code utilization in discrete data.
problem Inflexible decoders in VAEs lead to poor utilization of latent codes in discrete data.
method Proposed DD-VAE with deterministic decoding and new proposal distributions.
result DD-VAE improves latent code utilization and structure of learned manifold.
Conventional embedding methods directly associate each symbol with a continuous embedding vector, which is equivalent to applying a linear transformation based on a "one-hot" encoding of the discrete symbols. Despite its simplicity, such approach yields the number of parameters that grows linearly with the vocabulary s…
Paper proves method for calculating NML code length works for continuous models.
problem Uncertainty in calculating NML code length for continuous models.
method Introduced a novel decomposition approach based on the coarea formula to prove correctness for continuous cases.
result Method accurately calculates NML code length for continuous models.
FSDH is a fast supervised hashing method that outperforms existing methods.
problem Efficiently hashing large datasets for fast retrieval.
method FSDH uses a simple regression of class labels to hash codes, avoiding iterative discrete optimization.
result FSDH is significantly faster (12-151x) than existing methods like SDH and FastHash.
Adversarial attacks found to be effective on code models.
problem Adversarial robustness of code models.
method Instantiated adversarial attacks for code, demonstrated vulnerability, and improved robustness.
result Neural models for code are vulnerable to adversarial attacks.
IDF++ improves integer discrete flows for lossless compression.
problem Theoretical limitations of integer discrete flows for lossless compression.
method Investigated and improved integer discrete flows, addressing gradient bias and architecture modifications.
result Different architecture modifications improve integer discrete flows for lossless compression.
Unsupervised model generates distinct intonation codes for speech synthesis.
problem Lack of understanding of what prosodic variations are controlled in speech synthesis.
method Phrase-level variational autoencoder with multi-modal prior, using mode centres as intonation codes.
result Generated intonation codes are perceptually distinct and carry various affect-related styles.
New method improves image compression using bits-back coding.
problem Lossy image compression with deep latent variable models.
method Iterative inference, stochastic annealing, bits-back coding.
result New state-of-the-art performance on lossy image compression.
Paper proposes IABF to improve NECST robustness.
problem Limited robustness of NECST learned coding networks.
method Infomax Adversarial-Bit-Flip (IABF) to improve stability and robustness.
result IABF achieves state-of-the-art performances on compression and error correction benchmarks.
Proposes a framework to improve VAE latent codes using mutual information.
problem Lack of explicit measure for VAE latent variable quality.
method Variational Mutual Information Maximization Framework.
result Improves relationships between latent codes and observations.
We apply belief propagation to a Bayesian bipartite graph composed of discrete independent hidden variables and discrete visible variables. The network is the Discrete counterpart of Independent Component Analysis (DICA) and it is manipulated in a factor graph form for inference and learning. A full set of simulations …
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.
GMBL uses graph embedding to learn binary codes from multiple views for clustering.
problem Lack of complete structure and complementary information from multiple views in single-view hash clustering methods.
method Graph-based Multi-view Binary Learning (GMBL) using Laplacian matrix to preserve data structure and assign weights to views.
result GMBL outperforms previous methods in clustering performance on multiple datasets.
CLOUD method detects causal relationships in various data types without latent variable assumptions.
problem Detecting causal relationships in the presence of unobserved common causes.
method CLOUD method using Normalized Maximum Likelihood (NML) Code for various data types (discrete, mixed, continuous).
result CLOUD method is more effective than existing methods in inferring causal relationships.
New method fine-tunes discrete diffusion models for RLHF tasks.
problem Fine-tuning discrete diffusion models with policy gradient methods is challenging.
method Proposed SEPO algorithm for efficient fine-tuning over non-differentiable rewards.
result Numerical experiments show scalability and efficiency of SEPO.
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%.
We describe discrete restricted Boltzmann machines: probabilistic graphical models with bipartite interactions between visible and hidden discrete variables. Examples are binary restricted Boltzmann machines and discrete naive Bayes models. We detail the inference functions and distributed representations arising in th…
In this paper we consider the problem of transmitting a continuous alphabet discrete-time source over an AWGN channel. The design of good curves for this purpose relies on geometrical properties of spherical codes and projections of N-dimensional lattices. We propose a constructive scheme based on a set of curves on …
Proposes a framework to maximize mutual information in VAE models for better latent code representation.
problem Lack of explicit measurement of the quality of learned representations in VAE models.
method Variational Mutual Information Maximization Framework for VAE.
result Maximizes mutual information between latent codes and observations, improving latent code representation.
Unified framework for discrete diffusion modeling with flexible noising processes.
problem Efficient modeling of large discrete state spaces with arbitrary corruption dynamics.
method Generalized Discrete Diffusion from Snapshots (GDDS) framework that supports uniformization for fast noising and snapshot-based ELBO for reverse process.
result GDDS outperforms existing discrete diffusion methods in training efficiency and generation quality.
For reliable transmission across a noisy communication channel, classical results from information theory show that it is asymptotically optimal to separate out the source and channel coding processes. However, this decomposition can fall short in the finite bit-length regime, as it requires non-trivial tuning of hand-…
Unified discrete diffusion for categorical data simplifies training and sampling.
problem Training and sampling in discrete diffusion models for categorical data.
method Mathematical simplifications and elegant unification of discrete-time and continuous-time discrete diffusion.
result Unified Simplified Discrete Denoising Diffusion (USD3) outperforms SOTA baselines.
PRISM-VQ combines financial priors with vector quantization for better stock prediction.
problem Predicting cross-sectional stock returns is hard due to low signal-to-noise ratios and changing market conditions.
method Integrates expert priors, vector-quantized latent factors, and dynamic factor loadings.
result Consistent improvements in cross-sectional return prediction and portfolio performance.
A method for disentangling discrete and continuous factors of data without using a discriminator network.
problem Unsupervised disentanglement of discrete and continuous factors of data.
method A procedure that minimizes total correlation of continuous latent variables and a separate discrete inference procedure.
result The method significantly outperforms current disentanglement methods based on disentanglement score and inference network classification score.
Exact guidance for discrete data improves posterior sampling efficiency.
problem Inefficient guidance for discrete data in posterior sampling.
method Derive exact transition rate for desired distribution given learned discrete flow matching model.
result Significantly improved efficiency with single forward pass per sampling step.
New hashing method improves document retrieval precision.
problem Efficiently retrieving similar documents from large text databases.
method Pairwise supervised hashing with Bernoulli VAE and unbiased gradient estimator.
result Superior performance compared to existing methods.
MedGraph learns patient visit embeddings from EMRs, capturing both attributes and temporal sequences.
problem Limited EMR embedding methods fail to capture patient demographics, utilisation, and code descriptions.
method MedGraph constructs an attributed bipartite graph and uses a point process to model temporal sequences.
result MedGraph outperforms state-of-the-art methods in medical risk prediction tasks.
ADCMs adaptively discretize CMs for efficient training.
problem Manual discretization schemes cause repeated adjustments for different noise schedules and datasets.
method Unified framework with optimization problem, local and global consistency constraints, and Gauss-Newton method.
result Significantly improve training efficiency and generative performance of CMs.
CADD improves generative quality by augmenting discrete diffusion with continuous latent space.
problem Loss of semantic information between denoising steps in discrete diffusion models.
method Introduces a framework that augments discrete state space with a continuous latent space, allowing for graded, informative masked tokens.
result CADD improves generative quality across text generation, image synthesis, and code modeling.
ReDi improves few-step generation for discrete data models.
problem Slow sampling speeds in discrete flow-based models.
method Rectified Discrete Flow (ReDi) reduces factorization error by rectifying coupling.
result Empirically, ReDi reduces Conditional Total Correlation and enables few-step generation.
Deep neural network learns discrete state abstractions for efficient planning.
problem Efficient sequential decision making in large state spaces.
method Information bottleneck method for learning approximate bisimulations using deep neural encoders and action-conditioned HMM.
result Trained method efficiently plans for unseen goals in multi-goal reinforcement learning.
CANDI solves the gap between continuous and discrete diffusion models for text generation.
problem Underperformance of continuous diffusion models in discrete data domains.
method Introduces token identifiability and a hybrid framework (CANDI) to decouple discrete and continuous corruption.
result CANDI successfully avoids temporal dissonance, enabling continuous diffusion benefits for discrete spaces.
Tutorial on information bottleneck problems with connections to coding and learning.
problem Information bottleneck problems and their connections to coding and learning.
method Information theoretic perspective, practical methods, connections to various problems.
result Optimal trade-offs between relevance and complexity in discrete and vector Gaussian frameworks.
EB-GFN models discrete data with amortized MCMC sampling.
problem Probabilistic modeling of high-dimensional discrete data.
method EB-GFN combines GFlowNets with energy-based models for efficient sampling.
result EB-GFN effectively models various discrete data tasks.
Corrected samplers reduce discretization error in discrete flow models without additional computational cost.
problem Discretization error in samplers for discrete flow models.
method Established non-asymptotic error bounds for samplers, proposed time-corrected and location-corrected samplers.
result Location-corrected sampler has lower complexity and better generation quality.
Graph Beta Diffusion (GBD) generates graphs with mixed discrete and continuous components.
problem Generating graphs with mixed discrete and continuous components.
method Introduces Graph Beta Diffusion (GBD) using a beta diffusion process.
result Competes strongly with existing models across graph benchmarks.
Non-autoregressive method speeds up protein folding prediction 23 times.
problem Generating protein sequences with higher order interactions.
method Discrete diffusion conditioned on 3D structure using ProteinMPNN.
result 23 times speed up in inference without performance loss.
Deep generative models have been successfully used to learn representations for high-dimensional discrete spaces by representing discrete objects as sequences and employing powerful sequence-based deep models. Unfortunately, these sequence-based models often produce invalid sequences: sequences which do not represent a…
In supervised binary hashing, one wants to learn a function that maps a high-dimensional feature vector to a vector of binary codes, for application to fast image retrieval. This typically results in a difficult optimization problem, nonconvex and nonsmooth, because of the discrete variables involved. Much work has sim…
ProSper learns data components with non-standard priors and superpositions.
problem Learning complex data components with non-standard priors and superpositions.
method Probabilistic algorithms for sparse coding with non-standard priors and superpositions.
result Library supports scalable and parallelizable dictionary learning for large-scale applications.
Recurrent models for sequences have been recently successful at many tasks, especially for language modeling and machine translation. Nevertheless, it remains challenging to extract good representations from these models. For instance, even though language has a clear hierarchical structure going from characters throug…
The incredible variety of galaxy shapes cannot be summarized by human defined discrete classes of shapes without causing a possibly large loss of information. Dictionary learning and sparse coding allow us to reduce the high dimensional space of shapes into a manageable low dimensional continuous vector space. Statisti…
New JSA autoencoders tackle discrete latent variable models for semi-supervised learning.
problem Handling discrete observations and latent codes in deep generative models.
method Joint-stochastic-approximation (JSA) autoencoders that directly maximize data log-likelihood and minimize KL divergence.
result JSA autoencoders achieve comparable performance to continuous latent space models in semi-supervised tasks.