Recurrent neural networks (RNNs) in the brain and in silico excel at solving tasks with intricate temporal dependencies. Long timescales required for solving such tasks can arise from properties of individual neurons (single-neuron timescale, $\tau$, e.g., membrane time constant in biological neurons) or recurrent interactions among them (network-mediated timescale). However, the contribution of each mechanism for optimally solving memory-dependent tasks remains poorly understood. Here, we train RNNs to solve $N$-parity and $N$-delayed match-to-sample tasks with increasing memory requirements controlled by $N$ by simultaneously optimizing recurrent weights and $\tau$s. We find that for both tasks RNNs develop longer timescales with increasing $N$, but depending on the learning objective, they use different mechanisms. Two distinct curricula define learning objectives: sequential learning of a single-$N$ (single-head) or simultaneous learning of multiple $N$s (multi-head). Single-head networks increase their $\tau$ with $N$ and are able to solve tasks for large $N$, but they suffer from catastrophic forgetting. However, multi-head networks, which are explicitly required to hold multiple concurrent memories, keep $\tau$ constant and develop longer timescales through recurrent connectivity. Moreover, we show that the multi-head curriculum increases training speed and network stability to ablations and perturbations, and allows RNNs to generalize better to tasks beyond their training regime. This curriculum also significantly improves training GRUs and LSTMs for large-$N$ tasks. Our results suggest that adapting timescales to task requirements via recurrent interactions allows learning more complex objectives and improves the RNN's performance.
相關內容
We collect robust proposals given in the field of regression models with heteroscedastic errors. Our motivation stems from the fact that the practitioner frequently faces the confluence of two phenomena in the context of data analysis: non--linearity and heteroscedasticity. The impact of heteroscedasticity on the precision of the estimators is well--known, however the conjunction of these two phenomena makes handling outliers more difficult. An iterative procedure to estimate the parameters of a heteroscedastic non--linear model is considered. The studied estimators combine weighted $MM-$regression estimators, to control the impact of high leverage points, and a robust method to estimate the parameters of the variance function.
We combine the recent relaxation approach with multiderivative Runge-Kutta methods to preserve conservation or dissipation of entropy functionals for ordinary and partial differential equations. Relaxation methods are minor modifications of explicit and implicit schemes, requiring only the solution of a single scalar equation per time step in addition to the baseline scheme. We demonstrate the robustness of the resulting methods for a range of test problems including the 3D compressible Euler equations. In particular, we point out improved error growth rates for certain entropy-conservative problems including nonlinear dispersive wave equations.
Online learning holds the promise of enabling efficient long-term credit assignment in recurrent neural networks. However, current algorithms fall short of offline backpropagation by either not being scalable or failing to learn long-range dependencies. Here we present a high-performance online learning algorithm that merely doubles the memory and computational requirements of a single inference pass. We achieve this by leveraging independent recurrent modules in multi-layer networks, an architectural motif that has recently been shown to be particularly powerful. Experiments on synthetic memory problems and on the challenging long-range arena benchmark suite reveal that our algorithm performs competitively, establishing a new standard for what can be achieved through online learning. This ability to learn long-range dependencies offers a new perspective on learning in the brain and opens a promising avenue in neuromorphic computing.
Textual data from financial filings, e.g., the Management's Discussion \& Analysis (MDA) section in Form 10-K, has been used to improve the prediction accuracy of bankruptcy models. In practice, however, we cannot obtain the MDA section for all public companies. The two main reasons for the lack of MDA are: (i) not all companies are obliged to submit the MDA and (ii) technical problems arise when crawling and scrapping the MDA section. This research introduces for the first time, to the best of our knowledge, the concept of multimodal learning in bankruptcy prediction models to solve the problem that for some companies we are unable to obtain the MDA text. We use the Conditional Multimodal Discriminative (CMMD) model to learn multimodal representations that embed information from accounting, market, and textual modalities. The CMMD model needs a sample with all data modalities for model training. At test time, the CMMD model only needs access to accounting and market modalities to generate multimodal representations, which are further used to make bankruptcy predictions. This fact makes the use of bankruptcy prediction models using textual data realistic and possible, since accounting and market data are available for all companies unlike textual data. The empirical results in this research show that the classification performance of our proposed methodology is superior compared to that of a large number of traditional classifier models. We also show that our proposed methodology solves the limitation of previous bankruptcy models using textual data, as they can only make predictions for a small proportion of companies.
Imaging through perturbed multimode fibres based on deep learning has been widely researched. However, existing methods mainly use target-speckle pairs in different configurations. It is challenging to reconstruct targets without trained networks. In this paper, we propose a physics-assisted, unsupervised, learning-based fibre imaging scheme. The role of the physical prior is to simplify the mapping relationship between the speckle pattern and the target image, thereby reducing the computational complexity. The unsupervised network learns target features according to the optimized direction provided by the physical prior. Therefore, the reconstruction process of the online learning only requires a few speckle patterns and unpaired targets. The proposed scheme also increases the generalization ability of the learning-based method in perturbed multimode fibres. Our scheme has the potential to extend the application of multimode fibre imaging.
Validity problems in clinical machine learning by indirect data labeling using consensus definitions
We demonstrate a validity problem of machine learning in the vital application area of disease diagnosis in medicine. It arises when target labels in training data are determined by an indirect measurement, and the fundamental measurements needed to determine this indirect measurement are included in the input data representation. Machine learning models trained on this data will learn nothing else but to exactly reconstruct the known target definition. Such models show perfect performance on similarly constructed test data but will fail catastrophically on real-world examples where the defining fundamental measurements are not or only incompletely available. We present a general procedure allowing identification of problematic datasets and black-box machine learning models trained on them, and exemplify our detection procedure on the task of early prediction of sepsis.
This paper concerns the approximation of smooth, high-dimensional functions from limited samples using polynomials. This task lies at the heart of many applications in computational science and engineering - notably, some of those arising from parametric modelling and computational uncertainty quantification. It is common to use Monte Carlo sampling in such applications, so as not to succumb to the curse of dimensionality. However, it is well known that such a strategy is theoretically suboptimal. Specifically, there are many polynomial spaces of dimension $n$ for which the sample complexity scales log-quadratically, i.e., like $c \cdot n^2 \cdot \log(n)$ as $n \rightarrow \infty$. This well-documented phenomenon has led to a concerted effort over the last decade to design improved, and moreover, near-optimal strategies, whose sample complexities scale log-linearly, or even linearly in $n$. In this work we demonstrate that Monte Carlo is actually a perfectly good strategy in high dimensions, despite its apparent suboptimality. We first document this phenomenon empirically via a systematic set of numerical experiments. Next, we present a theoretical analysis that rigorously justifies this fact in the case of holomorphic functions of infinitely-many variables. We show that there is a least-squares approximation based on $m$ Monte Carlo samples whose error decays algebraically fast in $m/\log(m)$, with a rate that is the same as that of the best $n$-term polynomial approximation. This result is non-constructive, since it assumes knowledge of a suitable polynomial subspace in which to perform the approximation. We next present a compressed sensing-based scheme that achieves the same rate, except for a larger polylogarithmic factor. This scheme is practical, and numerically it performs as well as or better than well-known adaptive least-squares schemes.
This paper investigates the multiple testing problem for high-dimensional sparse binary sequences, motivated by the crowdsourcing problem in machine learning. We study the empirical Bayes approach for multiple testing on the high-dimensional Bernoulli model with a conjugate spike and uniform slab prior. We first show that the hard thresholding rule deduced from the posterior distribution is suboptimal. Consequently, the $\ell$-value procedure constructed using this posterior tends to be overly conservative in estimating the false discovery rate (FDR). We then propose two new procedures based on $\adj\ell$-values and $q$-values to correct this issue. Sharp frequentist theoretical results are obtained, demonstrating that both procedures can effectively control the FDR under sparsity. Numerical experiments are conducted to validate our theory in finite samples. To our best knowledge, this work provides the first uniform FDR control result in multiple testing for high-dimensional sparse binary data.
Clinical texts, such as admission notes, discharge summaries, and progress notes, contain rich and valuable information that can be used for various clinical outcome prediction tasks. However, applying large language models, such as BERT-based models, to clinical texts poses two major challenges: the limitation of input length and the diversity of data sources. This paper proposes a novel method to preserve the knowledge of long clinical texts using aggregated ensembles of large language models. Unlike previous studies which use model ensembling or text aggregation methods separately, we combine ensemble learning with text aggregation and train multiple large language models on two clinical outcome tasks: mortality prediction and length of stay prediction. We show that our method can achieve better results than baselines, ensembling, and aggregation individually, and can improve the performance of large language models while handling long inputs and diverse datasets. We conduct extensive experiments on the admission notes from the MIMIC-III clinical database by combining multiple unstructured and high-dimensional datasets, demonstrating our method's effectiveness and superiority over existing approaches. We also provide a comprehensive analysis and discussion of our results, highlighting our method's applications and limitations for future research in the domain of clinical healthcare. The results and analysis of this study is supportive of our method assisting in clinical healthcare systems by enabling clinical decision-making with robust performance overcoming the challenges of long text inputs and varied datasets.
Artificial neural networks thrive in solving the classification problem for a particular rigid task, acquiring knowledge through generalized learning behaviour from a distinct training phase. The resulting network resembles a static entity of knowledge, with endeavours to extend this knowledge without targeting the original task resulting in a catastrophic forgetting. Continual learning shifts this paradigm towards networks that can continually accumulate knowledge over different tasks without the need to retrain from scratch. We focus on task incremental classification, where tasks arrive sequentially and are delineated by clear boundaries. Our main contributions concern 1) a taxonomy and extensive overview of the state-of-the-art, 2) a novel framework to continually determine the stability-plasticity trade-off of the continual learner, 3) a comprehensive experimental comparison of 11 state-of-the-art continual learning methods and 4 baselines. We empirically scrutinize method strengths and weaknesses on three benchmarks, considering Tiny Imagenet and large-scale unbalanced iNaturalist and a sequence of recognition datasets. We study the influence of model capacity, weight decay and dropout regularization, and the order in which the tasks are presented, and qualitatively compare methods in terms of required memory, computation time, and storage.
小貼士
登錄享
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191