Abstract
Extrapolation of adverse biological (toxic) effects of chemicals is an important contribution to expand available hazard data in (eco)toxicology without the use of animals in laboratory experiments. In this work, we extrapolate effects based on a knowledge graph (KG) consisting of the most relevant effect data as domain-specific background knowledge. An effect prediction model, with and without background knowledge, was used to predict mean adverse biological effect concentration of chemicals as a prototypical type of stressors. The background knowledge improves the model prediction performance by up to 40\% in terms of $R^2$ (\ie coefficient of determination). We use the KG and KG embeddings to provide quantitative and qualitative insights into the predictions. These insights are expected to improve the confidence in effect prediction. Larger scale implementation of such extrapolation models should be expected to support hazard and risk assessment, by simplifying and reducing testing needs.
Towards Few-Shot Open-Set Object Detection
Authors: Binyi Su, Hua Zhang, Zhong Zhou
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Open-set object detection (OSOD) aims to detect the known categories and identify unknown objects in a dynamic world, which has achieved significant attentions. However, previous approaches only consider this problem in data-abundant conditions. We seek a solution for few-shot open-set object detection (FSOSOD), which aims to quickly train a detector based on few samples while detecting all known classes and identifying unknown classes. The main challenge for this task is that few training samples tend to overfit on the known classes, and lead to poor open-set performance. We propose a new FSOSOD algorithm to tackle this issue, named FOOD, which contains a novel class dropout cosine classifier (CDCC) and a novel unknown decoupling learner (UDL). To prevent over-fitting, CDCC randomly inactivates parts of the normalized neurons for the logit prediction of all classes, and then decreases the co-adaptability between the class and its neighbors. Alongside, UDL decouples training the unknown class and enables the model to form a compact unknown decision boundary. Thus, the unknown objects can be identified with a confidence probability without any pseudo-unknown samples for training. We compare our method with several state-of-the-art OSOD methods in few-shot scenes and observe that our method improves the recall of unknown classes by 5%-9% across all shots in VOC-COCO dataset setting.
Measuring the Confidence of Traffic Forecasting Models: Techniques, Experimental Comparison and Guidelines towards Their Actionability
Authors: Ibai Laña, Ignacio (Iñaki)Olabarrieta, Javier Del Ser
Abstract
The estimation of the amount of uncertainty featured by predictive machine learning models has acquired a great momentum in recent years. Uncertainty estimation provides the user with augmented information about the model's confidence in its predicted outcome. Despite the inherent utility of this information for the trustworthiness of the user, there is a thin consensus around the different types of uncertainty that one can gauge in machine learning models and the suitability of different techniques that can be used to quantify the uncertainty of a specific model. This subject is mostly non existent within the traffic modeling domain, even though the measurement of the confidence associated to traffic forecasts can favor significantly their actionability in practical traffic management systems. This work aims to cover this lack of research by reviewing different techniques and metrics of uncertainty available in the literature, and by critically discussing how confidence levels computed for traffic forecasting models can be helpful for researchers and practitioners working in this research area. To shed light with empirical evidence, this critical discussion is further informed by experimental results produced by different uncertainty estimation techniques over real traffic data collected in Madrid (Spain), rendering a general overview of the benefits and caveats of every technique, how they can be compared to each other, and how the measured uncertainty decreases depending on the amount, quality and diversity of data used to produce the forecasts.
Keyword: scaling
FerroX : A GPU-accelerated, 3D Phase-Field Simulation Framework for Modeling Ferroelectric Devices
Abstract
We present a massively parallel, 3D phase-field simulation framework for modeling ferro-electric materials based scalable logic devices. We self-consistently solve the time-dependent Ginzburg Landau (TDGL) equation for ferroelectric polarization, Poisson equation for electric potential, and charge equation for carrier densities in semiconductor regions. The algorithm is implemented using the AMReX software framework, which provides effective scalability on manycore and GPU-based supercomputing architectures. We demonstrate the performance of the algorithm with excellent scaling results on NERSC multicore and GPU systems, with a significant (15x) speedup on the GPU using a node-by-node comparison. We further demonstrate the applicability of the code in simulations of ferroelectric domain-wall induced negative capacitance (NC) effect in Metal-Ferroelectric-Insulator-Metal (MFIM) and Metal-Ferroelectric-Insulator-Semiconductor-Metal (MFISM) devices. The charge (Q) v.s. applied voltage (V) responses for these structures clearly indicates stabilized negative capacitance.
Confident Approximate Policy Iteration for Efficient Local Planning in $q^π$-realizable MDPs
Authors: Gellért Weisz, András György, Tadashi Kozuno, Csaba Szepesvári
Abstract
We consider approximate dynamic programming in $\gamma$-discounted Markov decision processes and apply it to approximate planning with linear value-function approximation. Our first contribution is a new variant of Approximate Policy Iteration (API), called Confident Approximate Policy Iteration (CAPI), which computes a deterministic stationary policy with an optimal error bound scaling linearly with the product of the effective horizon $H$ and the worst-case approximation error $\epsilon$ of the action-value functions of stationary policies. This improvement over API (whose error scales with $H^2$) comes at the price of an $H$-fold increase in memory cost. Unlike Scherrer and Lesner [2012], who recommended computing a non-stationary policy to achieve a similar improvement (with the same memory overhead), we are able to stick to stationary policies. This allows for our second contribution, the application of CAPI to planning with local access to a simulator and $d$-dimensional linear function approximation. As such, we design a planning algorithm that applies CAPI to obtain a sequence of policies with successively refined accuracies on a dynamically evolving set of states. The algorithm outputs an $\tilde O(\sqrt{d}H\epsilon)$-optimal policy after issuing $\tilde O(dH^4/\epsilon^2)$ queries to the simulator, simultaneously achieving the optimal accuracy bound and the best known query complexity bound, while earlier algorithms in the literature achieve only one of them. This query complexity is shown to be tight in all parameters except $H$. These improvements come at the expense of a mild (polynomial) increase in memory and computational costs of both the algorithm and its output policy.
Using Supervised Deep-Learning to Model Edge-FBG Shape Sensors
Authors: Samaneh Manavi Roodsari, Antal Huck-Horvath, Sara Freund, Azhar Zam, Georg Rauter, Wolfgang Schade, Philippe C. Cattin
Abstract
Continuum robots in robot-assisted minimally invasive surgeries provide adequate access to target anatomies that are not directly reachable through small incisions. Achieving precise and reliable motion control of such snake-like manipulators necessitates an accurate navigation system that requires no line-of-sight and is immune to electromagnetic noises. Fiber Bragg Grating (FBG) shape sensors, particularly edge-FBGs, are promising tools for this task. However, in edge-FBG sensors, the intensity ratio between Bragg wavelengths carries the strain information that can be affected by undesired bending-related phenomena, making standard characterization techniques less suitable for these sensors. We showed in our previous work that a deep learning model has the potential to extract the strain information from the full edge-FBG spectrum and accurately predict the sensor's shape. In this paper, we conduct a more thorough investigation to find a suitable architectural design with lower prediction errors. We use the Hyperband algorithm to search for optimal hyperparameters in two steps. First, we limit the search space to layer settings, where the best-performing configuration gets selected. Then, we modify the search space for tuning the training and loss calculation hyperparameters. We also analyze various data transformations on the input and output variables, as data rescaling can directly influence the model's performance. Moreover, we performed discriminative training using Siamese network architecture that employs two CNNs with identical parameters to learn similarity metrics between the spectra of similar target values. The best-performing network architecture among all evaluated configurations can predict the sensor's shape with a median tip error of 3.11 mm.
A Functional-Space Mean-Field Theory of Partially-Trained Three-Layer Neural Networks
Authors: Zhengdao Chen, Eric Vanden-Eijnden, Joan Bruna
Subjects: Machine Learning (cs.LG); Optimization and Control (math.OC); Probability (math.PR); Machine Learning (stat.ML)
Abstract
To understand the training dynamics of neural networks (NNs), prior studies have considered the infinite-width mean-field (MF) limit of two-layer NN, establishing theoretical guarantees of its convergence under gradient flow training as well as its approximation and generalization capabilities. In this work, we study the infinite-width limit of a type of three-layer NN model whose first layer is random and fixed. To define the limiting model rigorously, we generalize the MF theory of two-layer NNs by treating the neurons as belonging to functional spaces. Then, by writing the MF training dynamics as a kernel gradient flow with a time-varying kernel that remains positive-definite, we prove that its training loss in $L_2$ regression decays to zero at a linear rate. Furthermore, we define function spaces that include the solutions obtainable through the MF training dynamics and prove Rademacher complexity bounds for these spaces. Our theory accommodates different scaling choices of the model, resulting in two regimes of the MF limit that demonstrate distinctive behaviors while both exhibiting feature learning.
Keyword: calibration
Mitigating Health Disparities in EHR via Deconfounder
Authors: Zheng Liu, Xiaohan Li, Philip Yu
Subjects: Machine Learning (cs.LG); Computers and Society (cs.CY)
Abstract
Health disparities, or inequalities between different patient demographics, are becoming crucial in medical decision-making, especially in Electronic Health Record (EHR) predictive modeling. To ensure the fairness of sensitive attributes, conventional studies mainly adopt calibration or re-weighting methods to balance the performance on among different demographic groups. However, we argue that these methods have some limitations. First, these methods usually mean a trade-off between the model's performance and fairness. Second, many methods completely attribute unfairness to the data collection process, which lacks substantial evidence. In this paper, we provide an empirical study to discover the possibility of using deconfounder to address the disparity issue in healthcare. Our study can be summarized in two parts. The first part is a pilot study demonstrating the exacerbation of disparity when unobserved confounders exist. The second part proposed a novel framework, Parity Medical Deconfounder (PriMeD), to deal with the disparity issue in healthcare datasets. Inspired by the deconfounder theory, PriMeD adopts a Conditional Variational Autoencoder (CVAE) to learn latent factors (substitute confounders) for observational data, and extensive experiments are provided to show its effectiveness.
Comparison of Stereo Matching Algorithms for the Development of Disparity Map
Authors: Hamid Fsian, Vahid Mohammadi, Pierre Gouton, Saeid Minaei
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Stereo Matching is one of the classical problems in computer vision for the extraction of 3D information but still controversial for accuracy and processing costs. The use of matching techniques and cost functions is crucial in the development of the disparity map. This paper presents a comparative study of six different stereo matching algorithms including Block Matching (BM), Block Matching with Dynamic Programming (BMDP), Belief Propagation (BP), Gradient Feature Matching (GF), Histogram of Oriented Gradient (HOG), and the proposed method. Also three cost functions namely Mean Squared Error (MSE), Sum of Absolute Differences (SAD), Normalized Cross-Correlation (NCC) were used and compared. The stereo images used in this study were from the Middlebury Stereo Datasets provided with perfect and imperfect calibrations. Results show that the selection of matching function is quite important and also depends on the images properties. Results showed that the BP algorithm in most cases provided better results getting accuracies over 95%.
Stop Measuring Calibration When Humans Disagree
Authors: Joris Baan, Wilker Aziz, Barbara Plank, Raquel Fernandez
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Abstract
Calibration is a popular framework to evaluate whether a classifier knows when it does not know - i.e., its predictive probabilities are a good indication of how likely a prediction is to be correct. Correctness is commonly estimated against the human majority class. Recently, calibration to human majority has been measured on tasks where humans inherently disagree about which class applies. We show that measuring calibration to human majority given inherent disagreements is theoretically problematic, demonstrate this empirically on the ChaosNLI dataset, and derive several instance-level measures of calibration that capture key statistical properties of human judgements - class frequency, ranking and entropy.
Keyword: out of distribution detection
There is no result
Keyword: out-of-distribution detection
There is no result
Keyword: expected calibration error
There is no result
Keyword: overconfident
There is no result
Keyword: overconfidence
There is no result
Keyword: confidence
Understanding Adverse Biological Effect Predictions Using Knowledge Graphs
Towards Few-Shot Open-Set Object Detection
Measuring the Confidence of Traffic Forecasting Models: Techniques, Experimental Comparison and Guidelines towards Their Actionability
Keyword: scaling
FerroX : A GPU-accelerated, 3D Phase-Field Simulation Framework for Modeling Ferroelectric Devices
Confident Approximate Policy Iteration for Efficient Local Planning in $q^π$-realizable MDPs
Using Supervised Deep-Learning to Model Edge-FBG Shape Sensors
A Functional-Space Mean-Field Theory of Partially-Trained Three-Layer Neural Networks
Keyword: calibration
Mitigating Health Disparities in EHR via Deconfounder
Comparison of Stereo Matching Algorithms for the Development of Disparity Map
Stop Measuring Calibration When Humans Disagree