A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf
.
Filters
HELP: A Dataset for Identifying Shortcomings of Neural Models in Monotonicity Reasoning
[article]
2019
arXiv
pre-print
Since no large dataset has been developed for monotonicity reasoning, it is still unclear whether the main obstacle is the size of datasets or the model architectures themselves. ...
We add it to training data for the state-of-the-art neural models and evaluate them on test sets for monotonicity phenomena. ...
Acknowledgement We thank our three anonymous reviewers for helpful suggestions. ...
arXiv:1904.12166v1
fatcat:cqu5c55qxrg5jdsejz3xsfijfm
HELP: A Dataset for Identifying Shortcomings of Neural Models in Monotonicity Reasoning
2019
Proceedings of the Eighth Joint Conference on Lexical and Computational Semantics (*SEM 2019)
Since no large dataset has been developed for monotonicity reasoning, it is still unclear whether the main obstacle is the size of datasets or the model architectures themselves. ...
We add it to training data for the state-of-the-art neural models and evaluate them on test sets for monotonicity phenomena. ...
Acknowledgement We thank our three anonymous reviewers for helpful suggestions. ...
doi:10.18653/v1/s19-1027
dblp:conf/starsem/YanakaMBISAB19
fatcat:nitpb4ca2bgh3eomzp4ks6szma
Decomposing Natural Logic Inferences in Neural NLI
[article]
2021
arXiv
pre-print
In the interest of interpreting neural NLI models and their reasoning strategies, we carry out a systematic probing study which investigates whether these models capture the crucial semantic features central ...
Correctly identifying valid inferences in downward-monotone contexts is a known stumbling block for NLI performance, subsuming linguistic phenomena such as negation scope and generalized quantifiers. ...
The shortcomings of natural logic handling in various neural NLI models have been shown with several behavioural studies, where NLI challenge sets exhibiting examples of downward monotone reasoning are ...
arXiv:2112.08289v1
fatcat:zcrhzaaxcze6hgqkigq7vrgsje
MonoNet: Towards Interpretable Models by Learning Monotonic Features
[article]
2019
arXiv
pre-print
We argue that by enforcing monotonicity between features and outputs, we are able to reason about the effect of a single feature on an output independently from other features, and consequently better ...
called for a more active conversation towards a rigorous approach to interpretability. ...
This is needed for Goal 1. Expressiveness In order to identify biases in a dataset or in other models (Goal 2), the interpretable model should be unbiased, i.e. an universal approximator. ...
arXiv:1909.13611v1
fatcat:w5jgookpardlflw5r2c4hqsuli
With Greater Distance Comes Worse Performance: On the Perspective of Layer Utilization and Model Generalization
[article]
2022
arXiv
pre-print
Generalization of deep neural networks remains one of the main open problems in machine learning. ...
In this paper, we empirically examined how different layers of neural networks contribute differently to the model; we found that early layers generally learn representations relevant to performance on ...
For fully connected neural networks without skip layers, the contribution of neural networks layers follows a monotonic pattern, where early layers (closer to input) start taking effect early on, and deeper ...
arXiv:2201.11939v1
fatcat:pwmlbhb2ubaihelkdzwlnmzsj4
Periodic Spectral Ergodicity: A Complexity Measure for Deep Neural Networks and Neural Architecture Search
[article]
2020
arXiv
pre-print
We demonstrate the usefulness of this measure in practice on two sets of vision models, ResNet and VGG, and sketch the computation of cPSE for more complex network structures. ...
Because of this cascading approach, i.e., a symmetric divergence of PSE on the consecutive layers, it is possible to use this measure for Neural Architecture Search (NAS). ...
these datasets neatly. ...
arXiv:1911.07831v4
fatcat:nh6lfc6lcbfmpfv5tqoctepnvi
A neural network filtering approach for similarity-based remaining useful life estimation
2018
The International Journal of Advanced Manufacturing Technology
The framework proposed in this paper considers a similarity-based prognostic algorithm that is fed by the use of data normalisation and filtering methods for operational trajectories of complex systems ...
This is combined with a data-driven prognostic technique based on feed-forward neural networks with multi-regime normalisation. ...
The main reason for selection of this fitting approach is that the fitted HIs, y, for training trajectories have only increasing values and early stages are behaving in such a way that the fitting has ...
doi:10.1007/s00170-018-2874-0
fatcat:adgkgaydmvcknprlhbdcoc3xku
Statistics and data mining techniques for lifetime value modeling
1999
Proceedings of the fifth ACM SIGKDD international conference on Knowledge discovery and data mining - KDD '99
many of the shortcomings of each separate tool setresulting in a LTV tenure prediction model that is both accurate and understandable. ...
A new neural network model for hazard prediction is used to free proportional hazards-like models from their linearity and proportionality constraints, and clustering tools are applied to identify segments ...
. large datasets over a large number of time intervals, where the models need to be refreshed frequently with reasonable turn around times in a corporate setting. ...
doi:10.1145/312129.312205
dblp:conf/kdd/ManiDBD99
fatcat:lwzgmipuo5ccfk5ish3dsbgw2m
Sample-Efficient Reinforcement Learning via Counterfactual-Based Data Augmentation
[article]
2020
arXiv
pre-print
Reinforcement learning (RL) algorithms usually require a substantial amount of interaction data and perform well only for specific tasks in a fixed environment. ...
To address the issues of mechanism heterogeneity and related data scarcity, we propose a data-efficient RL algorithm that exploits structural causal models (SCMs) to model the state dynamics, which are ...
, we achieve flexible counterfactual reasoning in the general case, by adopting neural networks for function approximation of the SCM with no hard restrictions on data distribution or causal mechanisms ...
arXiv:2012.09092v1
fatcat:ejv3zbxnibcerkwyidm6i2l2xa
A frgsnn hybrid feature selection combining frgs filter and gsnn wrapper
2016
International Journal of Latest Trends in Engineering and Technology
How to selecting a small subset out of the thousands of genes in microarray data is important for accurate classification of phenotypes. ...
of the two specified models for gene selection. ...
For this reason, a mutation operator capable of spontaneously generating new chromosomes is included. ...
doi:10.21172/1.72.502
fatcat:anidioanyzctrajskojj2tltni
Cardiac Phase Detection in Echocardiograms With Densely Gated Recurrent Neural Networks and Global Extrema Loss
2019
IEEE Transactions on Medical Imaging
The proposed architectures integrate convolution neural networks (CNNs)-based image feature extraction model and recurrent neural networks (RNNs) to model temporal dependencies between each frame in a ...
The optimal deep learning model consists of a DenseNet and GRU trained with the proposed loss function. ...
ACKNOWLEDGMENT The authors would like to thank Dale Hawley for helping with accessing data used in this research. ...
doi:10.1109/tmi.2018.2888807
pmid:30582532
fatcat:yssmwjupkrdvbjlyd2gh74l5ju
Effective Sparsification of Neural Networks with Global Sparsity Constraint
[article]
2021
arXiv
pre-print
Weight pruning is an effective technique to reduce the model size and inference time for deep neural networks in real-world deployments. ...
As a by-product, we show ProbMask is also highly effective in identifying supermasks, which are subnetworks with high performance in a randomly weighted dense neural network. ...
ProbMask can also serve as a powerful tool for identifying subnetworks with high performance in a randomly weighted dense neural network.
A. ...
arXiv:2105.01571v1
fatcat:mdhmrmlyerbahmqgf2ju3mjyqu
Pushing the Limits of Low-Resource Morphological Inflection
[article]
2019
arXiv
pre-print
However, for a long tail of languages the necessary resources are hard to come by, and state-of-the-art neural methods that work well under higher resource settings perform poorly in the face of a paucity ...
Also, we identify the crucial factors for success with cross-lingual transfer for morphological inflection: typological similarity and a common representation across languages. ...
as to Gabriela Weigel for her invaluable help with editing and proofreading the paper. ...
arXiv:1908.05838v2
fatcat:ahvfqej45ngadixgkje46fzxra
Interpretable machine learning: Fundamental principles and 10 grand challenges
2022
Statistics Survey
better interpretability; (4) Modern case-based reasoning, including neural networks and matching for causal inference; (5) Complete supervised disentanglement of neural networks; (6) Complete or even partial ...
This survey is suitable as a starting point for statisticians and computer scientists interested in working in interpretable machine learning. ...
Thank you to the anonymous reviewers that made extremely helpful comments. ...
doi:10.1214/21-ss133
fatcat:ahzfoilhmfa2rd4hcauvsn3eyy
Capturing and incorporating expert knowledge into machine learning models for quality prediction in manufacturing
[article]
2022
arXiv
pre-print
In particular, the direct involvement of process experts in the training of the models leads to a very clear interpretation and, by extension, to a high acceptance of the models. ...
In manufacturing, the size of available datasets before start of production is often limited. In contrast to data, expert knowledge commonly is available in manufacturing. ...
Acknowledgments We gratefully acknowledge the funding provided by the Fraunhofer Society as part of the lighthouse project "Machine Learning for Production" (ML4P). ...
arXiv:2202.02003v1
fatcat:vhsjy5a75vewjg23nypz6o2mti
« Previous
Showing results 1 — 15 out of 2,120 results