We present PredRNN++, an improved recurrent network for video predictive learning. In pursuit of a greater spatiotemporal modeling capability, our approach increases the transition depth between adjacent states by leveraging a novel recurrent unit, which is named Causal LSTM for re-organizing the spatial and temporal m…
We provide high-probability sample complexity guarantees for exact structure recovery and accurate predictive learning using noise-corrupted samples from an acyclic (tree-shaped) graphical model. The hidden variables follow a tree-structured Ising model distribution, whereas the observable variables are generated by a …
TIDBD adapts step sizes online for better robotic predictions.
problem Choosing appropriate learning parameters for online prediction-learning.
method Temporal-Difference Incremental Delta-Bar-Delta (TIDBD) for step-size adaptation.
result TIDBD performs comparably to classic TD learning and detects sensor failures.
Algorithm constructs confidence sets for deep neural networks with PAC guarantees.
problem Ensuring reliable predictions for deep neural networks with high confidence.
method Combines calibrated prediction and learning theory bounds.
result Constructs PAC confidence sets for various deep models.
We use the language of uninformative Bayesian prior choice to study the selection of appropriately simple effective models. We advocate for the prior which maximizes the mutual information between parameters and predictions, learning as much as possible from limited data. When many parameters are poorly constrained by …
This paper introduces a new large-scale music dataset, MusicNet, to serve as a source of supervision and evaluation of machine learning methods for music research. MusicNet consists of hundreds of freely-licensed classical music recordings by 10 composers, written for 11 instruments, together with instrument/note annot…
Improves supervised learning with target-embedding autoencoders.
problem Improving generalization in purely supervised settings with high-dimensional target spaces.
method Target-Embedding Autoencoders (TEA) for jointly optimizing latent representations for prediction and feature predictability.
result Guaranteed generalization for linear TEAs through uniform stability, and empirical validation across multivariate sequence forecasting.
Predicts EHR features that improve prediction, learning coherent topics.
problem Balancing prediction quality and coherence in EHR topic models.
method Prediction-focused topic model that uses supervisory signal to retain relevant features.
result Prediction-focused topic models learn more coherent topics while maintaining competitive predictions.
Paper proposes a hybrid model for financial time series prediction using sentiment analysis.
problem Challenges in forecasting in non-stationary, complex environments with heterogeneous data.
method Hybrid model combining GANs with NLP-based sentiment analysis.
result Hybrid model enhances robustness in non-stationary environments.
A Bayesian network is a widely used probabilistic graphical model with applications in knowledge discovery and prediction. Learning a Bayesian network (BN) from data can be cast as an optimization problem using the well-known score-and-search approach. However, selecting a single model (i.e., the best scoring BN) can b…
AI learns to learn sequentially without forgetting.
problem Preventing catastrophic forgetting in machine learning models.
method Meta-learning a neuromodulatory activation-gating function to control selective activation in deep neural networks.
result State-of-the-art continual learning performance with 600 classes (9,000 updates).
Bayesian optimization finds best hyperparameters for deep learning models.
problem Finding optimal hyperparameters in high-dimensional space.
method Bayesian optimization, Gaussian processes, and adaptive experimentation.
result Ax, BoTorch, and GPyTorch provide a powerful and simple framework for hyperparameter optimization.
New study shows how model complexity affects test risk, challenging classical theory.
problem Understanding how test risk scales with model complexity for large over-parametrized deep networks.
method Developed norm-based capacity measures for random features based estimators, providing precise characterization of estimator's norm concentration and test error.
result Predicted learning curve shows a phase transition from under- to over-parameterization, confirming classical U-shaped behavior with appropriate capacity measures.
CloudLSTM forecasts geospatial point-cloud data streams accurately.
problem Forecasting over geospatial point-cloud data streams.
method Introduces CloudLSTM, a recurrent neural model with a Dynamic Point-cloud Convolution (DConv) operator.
result CloudLSTM outperforms competitor models in long-term predictions for point-cloud data.
Unified study of stateful replay for streaming learning, reducing forgetting by 2-3x.
problem Catastrophic forgetting in streaming generative and predictive learning.
method Unified analysis of stateful replay for autoencoding, forecasting, and classification tasks.
result Stateful replay reduces average forgetting by a factor of 2-3 on heterogeneous multi-task streams.
Paper proposes a method to make learned models focus on task-relevant information.
problem Mismatch between model objective and downstream task objective.
method Direct prediction towards task-relevant information, self-supervised.
result Model more effectively models relevant parts of the scene conditioned on the goal.
One of the current challenges in machine learning is how to deal with data coming at increasing rates in data streams. New predictive learning strategies are needed to cope with the high throughput data and concept drift. One of the data stream mining tasks where new learning strategies are needed is multi-target regre…
Weibo experts predict stock market better than non-experts.
problem Improving stock market prediction accuracy using sentiment analysis.
method Combining BERT for sentiment classification and LSTM for time-series prediction on Weibo data.
result AFA group users' predictions are 39.67% more accurate than UFA group users.
Building a good predictive model requires an array of activities such as data imputation, feature transformations, estimator selection, hyper-parameter search and ensemble construction. Given the large, complex and heterogenous space of options, off-the-shelf optimization methods are infeasible for realistic response t…
This paper analyzes self-supervised learning from a multi-view perspective.
problem Understanding and optimizing self-supervised learning from multi-view data.
method Information-theoretical framework to understand and design self-supervised learning objectives.
result Self-supervised representations can extract task-relevant information and discard task-irrelevant information.
Paper proposes a cost-sensitive conformal training method with provably controllable learning bounds.
problem Uncertainty quantification and learning bounds in conformal prediction.
method Cost-sensitive conformal training algorithm that minimizes the expected size of prediction sets using rank weighting.
result Theoretical analysis shows tightness between weighted objective and expected size of conformal prediction sets.
New algorithm for learning functions with bounds on error and sample complexity.
problem Learning [0,1]-valued functions in a prediction model. method General-purpose algorithm with upper and lower bounds on expected error and sample complexity.
result Improved bounds on sample complexity and agnostic learning conditions.
Many AI problems, in robotics and other domains, are goal-directed, essentially seeking a trajectory leading to some goal state. In such problems, the way we choose to represent a trajectory underlies algorithms for trajectory prediction and optimization. Interestingly, most all prior work in imitation and reinforcemen…
RegimeFolio optimizes portfolios by adapting to changing market regimes.
problem Non-stationary markets with shifting volatility regimes.
method Explicitly models volatility regimes with sector-specific ensemble forecasting and adaptive mean-variance allocation.
result Significant improvement in return and robustness compared to conventional methods.
Analyzes symmetries in neural networks to predict learning dynamics.
problem Understanding the dynamics of neural network parameters during training.
method Unified theoretical framework based on symmetries and conservation laws.
result Symmetries impose geometric constraints on gradients and Hessians, leading to conservation laws.
Paper develops a new model for forecasting ocean currents.
problem Forecasting the Loop Current and its eddies for the Gulf of Mexico.
method Physics-informed Tensor-train ConvLSTM, incorporating prior physical knowledge.
result PITT-ConvLSTM outperforms state-of-the-art methods in volumetric velocity forecasting.
LC-PFN predicts learning curve performance more accurately and faster than MCMC.
problem Bayesian extrapolation of learning curves is computationally expensive and overly restrictive.
method Prior-Data Fitted Neural Networks (PFNs) for approximate Bayesian inference.
result LC-PFN outperforms MCMC in accuracy and is significantly faster.
With the fast development of various positioning techniques such as Global Position System (GPS), mobile devices and remote sensing, spatio-temporal data has become increasingly available nowadays. Mining valuable knowledge from spatio-temporal data is critically important to many real world applications including huma…
Single model estimates ensemble uncertainty efficiently.
problem Efficient uncertainty quantification in deep learning models.
method Contextual similarity distillation, approximating ensemble variance with a single model.
result Single model estimates predictive variance with a single forward pass.
This paper investigates the problem of online prediction learning, where learning proceeds continuously as the agent interacts with an environment. The predictions made by the agent are contingent on a particular way of behaving, represented as a value function. However, the behavior used to select actions and generate…
Analysis of deep neural networks under various learning rules reveals dynamics of feature and prediction learning.
problem Understanding how different learning rules affect feature and prediction dynamics in deep neural networks.
method Analysis of infinite-width deep networks trained with gradient descent and various learning rules.
result The evolution of the output function is governed by an effective neural tangent kernel (eNTK), which varies depending on the learning rule and training regime.