Category: conformal prediction

Generalization and Informativeness of Conformal Prediction


When using a machine learning model to make important decisions, like in healthcare, finance, or engineering, we not only need accurate predictions but also want to know how sure the model is about its answers [1-3]. CP offers a practical solution for generating certified “error bars”—certified ranges of uncertainty—by post-processing the outputs of a fixed, pre-trained base predictor. This is crucial for safety and reliability. At the upcoming ISIT 2024 conference, we will present our research work, which aims to bridge the generalization properties of the base predictor with the expected size of the set predictions, also known as informativeness, produced by CP. Understanding the informativeness of CP is particularly relevant as it can usually only be assessed at test time.

Conformal prediction

Figure 1: Conformal prediction (CP) set predictors (gray areas) obtained by calibrating a base predictor with a higher generalization error on the left and a lower generalization error on the right. Thanks to CP, both set predictors satisfy a user-defined coverage guarantee, but the inefficiency, i.e., the average prediction set size, is larger when the generalization error of the base predictor is larger.

The most practical form of CP, known as inductive CP, divides the available data into a training set and a calibration set [4]. We use the training data to train a base model, and the calibration data to determine the prediction sets around the decisions made by the base model. As shown in Figure 1, a more accurate base predictor, which generalizes better outside the training set, tends to produce more informative sets when CP is applied.


Figure 2: Bound on the average set size for different values of training and calibration data set sizes as a function of the target reliability level. Increasing the number of calibration data points causes the bound to converge exponentially fast to a function (black line) that is increasing in and decreasing in the amount of training data.

Our work’s main contribution is a high probability bound on the expected size of the predicted sets. The bound relates the informativeness of CP to the generalization properties of the base model and the amount of available training and calibration data. As illustrated in Figure 2, our bound predicts that by increasing the amount of calibration data CP’s efficiency converges rapidly to a quantity influenced by the coverage level, the size of the training set, and the predictor’s generalization performance. However, for finite amount of calibration data, the bound is also influenced by the discrepancy between the target and empirical reliability measured over the training data set. Overall, the bound justifies a common practice: allocating more data to train the base model compared to the data used to calibrate it.

Figure 3: Normalized empirical CP set size for a multi-class classification problem on the MNIST data set as a function of the reliability level and for different sizes of the calibration and training data sets.

Since what really proves the worth of a theory is how well it holds up in real-world testing, we also compare our theoretical findings with numerical evaluations. In our study, we looked at two classification and regression tasks. We ran CP with various splits of calibration and training data, then measured the average efficiency. As shown in the Figure 3, the empirical results from our experiments matched up nicely with what our theory predicted in Figure 2.


[1] A. L. Beam and I. S. Kohane, “Big data and machine learning in health care,” JAMA, vol. 319, no. 13, pp. 1317–1318, 2018.

[2] J.. W. Goodell, S. Kumar, W. M. Lim, and D. Pattnaik, “Artificial intelligence and machine learning in finance: Identifying foundations, themes, and research clusters from bibliometric analysis,” Journal of Behavioral and Experimental Finance, vol. 32, p. 100577, 2021.

[3] L. Hewing, K. P. Wabersich, M. Menner, and M. N. Zeilinger, “Learning-based model predictive control: Toward safe learning in control,” Annual Review of Control, Robotics, and Autonomous Systems, vol. 3, pp. 269–296, 2020.

[4] V. Vovk, A. Gammerman, and G. Shafer, Algorithmic learning in a random world, vol. 29. Springer, 2005.

Wireless Reliable Federated Inference

Written by Meiyi Zhu during her visit to KCLIP.


Consider a wireless federated inference scenario in which the devices and a server share a pre-trained machine learning model, e.g., trained via federated learning. The server wishes to make an inference on its own new input based on such a pre-trained machine learning model. Note that the server has no access to the data; the data is only presented at the devices. This scenario is common in practice. For example, a personal healthcare system would first train the respective model via federated learning, without acquiring personal data from the end users; while upon achieving a trained healthcare model, wishes to provide useful solution to new users. We will assume that new users ask queries to the central server, while the general conclusion made in this article retains even for the case in which the new user has its own access to the pre-trained model.

However, depending on the quality of the pre-trained model, e.g., lack of data, the solution provided by the pre-trained model may yield wrong decisions. More importantly, such model is likely to yield unreliable decisions; see, e.g., our previous post ‘Is Accuracy Sufficient for AI in 6G? (No, Calibration is Equally Important)’. As reliability plays an important role in various fields including healthcare monitoring and autonomous vehicle navigation, it is important to find ways to make the federated inference reliable. But how can we make the pre-trained model reliable as the central server has no access to the data at all?

Recent work has introduced federated conformal prediction (CP), which improves the reliability of the server’s decision by utilizing available held-out local data at each device, of course, without central server’s access to such data. The goal of federated CP is to provide a guaranteed interval or set of potential outputs that contains the correct answer at a predefined reliability level [1, 2]. As a state-of-the-art solution, reference [1] proposed a quantile-of-quantile (QQ) scheme, referred to as FedCP-QQ, whereby each device computes and communicates a pre-determined quantile of the local losses. However, existing work assumed noise-free communication between the server and the devices, whereby devices can communicate a single real number to the server.

Wireless Federated Conformal Prediction

In our recent work, to appear in Transactions on Signal Processing, we study for the first time federated CP in a wireless setting, as illustrated in Fig. 1. Specifically, we introduce a novel protocol, termed wireless federated conformal prediction (WFCP), which builds on type-based multiple access (TBMA) and on a novel quantile correction scheme.

Fig. 1. Illustration of the wireless reliable federated inference problem under study.

TBMA is a multiple access scheme that aims at recovering aggregated statistics, rather than individual messages [3]. By noting that federated CP also requires aggregated statistics across the devices, i.e., quantile, we have proposed to apply TBMA for WFCP. More precisely, as illustrated in Fig. 2, TBMA enables the estimate of the global histogram of data available across all devices without having to separately estimate the histograms of all devices. Specifically, each histogram bin is assigned an orthogonal codeword and the server can estimate the global histogram thanks to the superposition property of wireless communications. In this way, WFCP enables a direct estimate of the global quantile at the server without imposing bandwidth requirements that scale linearly with the number of active devices like FedCP-QQ. Rather, the communication requirements of WFCP are only dictated by the precision with which the signals are represented for transmission to the server, i.e., the length of each codeword.

Fig. 2. Illustration of the TBMA enabled communication model.

The other key technical challenge tackled in our work is the derivation of a novel quantile correction approach that ensures the reliability of the set predictor despite the presence of channel noise.


We evaluate our proposed WFCP on CIFAR-10 data set over Rayleigh fading channels. We show here one of the results that plots the performance gains of WFCP in the presence of limited communication resources. In Fig. 3, we evaluate the performance of WFCP and our implementation of existing FedCP-QQ (DQQ) over wireless channels using finite blocklength information theory as a function of SNR. As SNR increases, both WFCP and DQQ maintain the target reliability level, while offering a decreasing prediction set size. Across all the SNRs, WFCP generates a more informative predicted set than DQQ, and it approaches the performance of the centralized CP. Please refer to our paper for more details.


Fig. 3. Empirical coverage and normalized empirical inefficiency of centralized CP, WFCP, and digital implementation of existing FedCP-QQ [1].


[1] P. Humbert, B. Le Bars, A. Bellet, and S. Arlot, “One-shot federated conformal prediction,” ICML 2023

[2] C. Lu and J. Kalpathy-Cramer, “Distribution-free federated learning with conformal predictions,” arXiv:2110.07661, 2021

[3 G. Mergen and L. Tong, “Type based estimation over multiaccess channels,” IEEE TSP 2006

Safe Model Predictive Control via Reliable Time-Series Forecasting


The control of dynamical systems is the backbone of modern technologies, ranging from industrial processes to autonomous vehicles. In many of these scenarios, systems must be controlled while satisfying a set of safety and reliability constraints with respect to the unknown evolution of a target process. For example, as illustrated in Figure 1, autonomous vehicles or unmanned aerial vehicles (UAVs) must plan their trajectory while maintaining a safe distance from other vehicles or obstacles. To this end, predictions about the future evolution of the system must be used. In this context, a primary challenge is to ensure safety and reliability in the face of predictions that are often uncertain.

Figure 1: UAV tracking problem, an example of model predictive control in which the UAV must plan its path based on the unknown evolution of the object to be tracked.

Probabilistic Time Series-Conformal Risk Prediction

To support the deployment of reliable control mechanisms for dynamical system, in our work we have recently proposed probabilistic time series-conformal risk prediction (PTS-CRC). PTS-CRC is a novel post-hoc calibration procedure that operates on the predictions produced by any pre-designed probabilistic forecaster to yield reliable time series prediction sets. As illustrated in Figure 2, PTS-CRC generates predictive sets based on an ensemble of multiple prototype trajectories sampled from the probabilistic model, supporting the efficient representation of forking uncertainties. This contrasts with previous solutions that apply Conformal Prediction[1] to deterministic predictors (TS-CP)[2], which are bounded to produce compact prediction sets. Furthermore, sets produced by PTS-CRC can be calibrated to satisfy a wide array of reliability definitions, beyond the standard one of coverage.

Figure 2: Construction of a prototype-based set predictor based on 3 prototypical sequences.

PTS-CRC Based Model Predictive Control

Based on the reliability properties of PTS-CRC predictions, we devise a novel Model Predictive Control (MPC) framework that addresses open-loop and closed-loop control problems under general average constraints on the quality or safety of the control policy. The key idea is to derive the control by replacing constraints that depend on the unknown dynamics of the target process with those depending on the predictive sets output by PTS-CRC. The reliability requirements of PTS-CRC predictions translate into reliability requirements for the original control problem.


We apply PTS-CRC-based MPC to wireless networking problems, specifically focusing on a scenario where a base station must modulate its future power allocation based on the unknown evolution of channel conditions. For instance, we address the challenge of controlling transmit power to maximize the communication rate at an unlicensed user while adhering to a safety requirement, expressed as the maximum interference experienced by a licensed user. By employing PTS-CRC, we can replace the unknown system evolution with efficient multimodal predictive sets that more effectively capture multimodal channel evolution compared to TS-CP (Figure 3). As exemplified in Figure 4, PTS-CRC-based power control leads to power allocations that achieve a higher communication rate compared to TS-CP.

Figure 3: Comparison between the prediction sets of TS-CP and PTS-CRC for the problem of channel gain evolution forecasting.

Figure 4: Comparison between the power control solution obtained using PTS-CRC and TS-CP based MPC.


[1] Vovk, Vladimir, Alexander Gammerman, and Glenn Shafer. “Algorithmic learning in a random world,” Vol. 29. New York: Springer, 2005.

[2] Stankeviciute, Kamile, Ahmed M Alaa, and Mihaela van der Schaar. “Conformal time-series forecasting.” Advances in neural information processing systems 34, 2021.

[3] Zecchin, Matteo, Sangwoo Park, and Osvaldo Simeone. “Forking Uncertainties: Reliable Prediction and Model Predictive Control with Sequence Models via Conformal Risk Control.” arXiv preprint arXiv:2310.10299, 2023.

Learning to Learn How to Calibrate

As discussed in our previous post ‘Is Accuracy Sufficient for AI in 6G? (No, Calibration is Equally Important)’, reliable AI should be able to quantify its uncertainty, i.e., to “know when it knows” and “know when it does not know”. To obtain reliable, or well-calibrated, AI models, two types of approaches can be adopted: (i) training-based calibration, and (ii) post-hoc calibration. Training-based calibration modifies the training procedure by accounting for calibration performance, and includes methods such as Bayesian learning [1, 2], robust Bayesian learning [3, 4], and calibration-aware regularization [5]; while post-hoc calibration utilizes validation data to “recalibrate” a probabilistic model, as in temperature scaling [6], Platt scaling [7], and isotonic regression [8]. All these methods have no formal guarantees on calibration, either due to inevitable model misspecification [9], or due to overfitting to the validation set [10, 11]. In contrast, conformal prediction (CP) offers formal calibration guarantees, although calibration is defined in terms of set, rather than probabilistic, prediction [12]. 

Fig. 1. Improvements in calibration can be obtained by either (i) training-based calibration or (ii) post-hoc calibration. Only conformal prediction, a post-hoc calibration approach, provides formal guarantees on calibration via set prediction.

A well-calibrated set predictor is the one that contains the true label with probability no smaller than a predetermined coverage level, say 90%. A set predictor obtained by conformal prediction is provably well calibrated, irrespective of the unknown underlying ground-truth distribution as long as the data examples are exchangeable, or i.i.d. (independent and identically distributed). 

One could trivially build a well-calibrated set predictor by producing the entire label set as the predicted set. However, such set predictor would be completely uninformative, since the size of the set predictor determines how informative the set predictor is. While conformal prediction is always guaranteed to yield reliable set predictors, it may produce large predicted set size in the presence of limited data examples [13]. In our recent work, presented at the NeurIPS 2022 Workshop on Meta-Learning, we have introduced a novel method that enhances the informativeness of CP-based set predictors via meta-learning.

Fig. 2. Meta-learning transfers knowledge from multiple tasks. In our recent paper, we have proposed an application of meta-learning to conformal prediction with the aim of reducing the average prediction set size while preserving formal calibration guarantees.

Meta-learning, or learning to learn, transfers knowledge from multiple tasks to optimize the inductive bias (e.g., the model class) for new, related, tasks [14]. In our recent work, meta-learning was applied to cross-validation-based conformal prediction (XB-CP) [13] to achieve well-calibrated and informative set predictors. As demonstrated in the following figure, the proposed meta-learning approach for XB-CP, termed meta-XB, can reduce the average prediction set size as compared to conventional CP approaches (XB-CP and validation-based conformal prediction (VB-CP) [12]) and to previous work on meta-learning for VB-CP [14], while preserving the formal guarantees on reliability (the predetermined coverage level, 90%, is always satisfied for meta-XB). 

Fig. 3. Average prediction set size (left) and coverage (right) for new tasks as a function of number of meta-training tasks. As compared to conventional CP schemes (VB-CP and XB-CP), meta-learning based approaches (meta-VB and meta-XB) have smaller prediction set size; while the proposed meta-XB guarantees reliability for every task unlike meta-VB that satisfies coverage condition on average over multiple tasks.

For more details including improvements in terms of input-conditional coverage via meta-learning with adaptive nonconformity scores [15], and further experimental results on image classification and communication engineering aspects, please refer to the arXiv posting.


[1] O. Simeone, Machine learning for engineers. Cambridge University Press, 2022

[2] J. Knoblauch, et al, “Generalized variational inference: Three arguments for deriving new posteriors,” arXiv:1904.02063, 2019

[3] W. Morningstar, et al “PACm-Bayes: Narrowing the empirical risk gap in the Misspecified Bayesian Regime,” NeurIPS 2021

[4] M. Zecchin, et al, “Robust PACm: Training ensemble models under model misspecification and outliers,” arXiv:2203.01859, 2022

[5] A. Kumar, et al, “Trainable calibration measures for neural networks from kernel mean embeddings,” ICML 2018

[6] C. Guo, et al, “On calibration of modern neural networks,” ICML 2017

[7] J. Platt, et al, “Probabilistic outputs for support vector machines and comparisons to regularized likelihood method,” Advances in Large Margin Classifiers 1999

[8]  B. Zadrozny and C. Elkan “Transforming classifier scores into accurate multiclass probability estimates,” KDD 2022

[9] A. Masegosa, “Learning under model misspecification: Applications to variational and ensemble methods.” NeurIPS 2020

[10] A. Kumar, et al, “Verified Uncertainty Calibration,” NeurIPS 2019

[11] X. Ma and M. B. Blaschko, “Meta-Cal: Well-controlled Post-hoc Calibration by Ranking,” ICML 2021 

[12]  V. Vovk, et al, “Algorithmic Learning in a Random World,” Springer 2005

[13] R. F. Barber, et al, “Predictive inference with the jackknife+,” The Annals of Statistics, 2021

[14] Chen, Lisha, et al. “Learning with limited samples—Meta-learning and applications to communication systems.” arXiv preprint arXiv:2210.02515, 2022.

[14] A. Fisch, et al, “Few-shot conformal prediction with auxiliary tasks,” ICML 2021

[15] Y. Romano, et al, “Classification with valid and adaptive coverage,” NeurIPS 2020