A central goal of explainable artificial intelligence (XAI) is to improve the trust relationship in human-AI interaction. One assumption underlying research in transparent AI systems is that explanations help to better assess predictions of machine learning (ML) models, for instance by enabling humans to identify wrong predictions more efficiently. Recent empirical evidence however shows that explanations can have the opposite effect: When presenting explanations of ML predictions humans often tend to trust ML predictions even when these are wrong. Experimental evidence suggests that this effect can be attributed to how intuitive, or human, an AI or explanation appears. This effect challenges the very goal of XAI and implies that responsible usage of transparent AI methods has to consider the ability of humans to distinguish machine generated from human explanations. Here we propose a quantitative metric for XAI methods based on Turing's imitation game, a Turing Test for Transparency. A human interrogator is asked to judge whether an explanation was generated by a human or by an XAI method. Explanations of XAI methods that can not be detected by humans above chance performance in this binary classification task are passing the test. Detecting such explanations is a requirement for assessing and calibrating the trust relationship in human-AI interaction. We present experimental results on a crowd-sourced text classification task demonstrating that even for basic ML models and XAI approaches most participants were not able to differentiate human from machine generated explanations. We discuss ethical and practical implications of our results for applications of transparent ML.

Recently introduced self-supervised methods for image representation learning provide on par or superior results to their fully supervised competitors, yet the corresponding efforts to explain the self-supervised approaches lag behind. Motivated by this observation, we introduce a novel visual probing framework for explaining the self-supervised models by leveraging probing tasks employed previously in natural language processing. The probing tasks require knowledge about semantic relationships between image parts. Hence, we propose a systematic approach to obtain analogs of natural language in vision, such as visual words, context, and taxonomy. Our proposal is grounded in Marr's computational theory of vision and concerns features like textures, shapes, and lines. We show the effectiveness and applicability of those analogs in the context of explaining self-supervised representations. Our key findings emphasize that relations between language and vision can serve as an effective yet intuitive tool for discovering how machine learning models work, independently of data modality. Our work opens a plethora of research pathways towards more explainable and transparent AI.

Machine Learning is the path to a better and advanced future. A Machine Learning Developer is the most demanding job in 2021 and it is going to increase by 20–30% in the upcoming 3–5 years. Machine Learning by the core is all statistics and programming concepts. The language that is mostly used by Machine learning developers for coding is python because of its simplicity. In this blog, you will some of the most asked machine learning questions that every machine learning enthusiast has to answer one day.

In addition to this, the recent'Big Bang' in large datasets across companies, organisation, and government departments has resulted in a large uptake in data mining techniques. So, what is data mining? Simply put, it's the process of discovering trends and insights in high-dimensionality datasets (those with thousands of columns). On the one hand, the high-dimensionality datasets have enabled organisations to solve complex, real-world problems, such as reducing cancer patient waiting time, predicting protein structure associated with COVID-19, and analysing MEG brain imaging scans. However, on the other hand, large datasets can sometimes contain columns with poor-quality data, which can lower the performance of the model -- more isn't always better.

I remember sitting in my 8th grade English class as we were all going around one day, naming a family member for whom we were grateful. I remember the boy who raised his hand shyly and shared that he was grateful for his mom, who had been battling breast cancer. After he shared his mother's story, multiple other people shared the name of their mothers, aunts, grandmothers, and close family members that had breast cancer. It was an eye-opening experience. Until then, I had not realized how prevalent breast cancer was.

Machine learning is gradually becoming critical part of life. From recommending movies to self driving cars, AI is making its presence felt in all walks of life. As ML models are taking critical decision, gradual need was felt to explain the decision taken by these models. Most of these model tend to be black box. While accurate perdition helps, answer to'why it was decided the way it was' is equally important .

Differential privacy (DP) Dwork et al. (2006b), Dwork et al. (2006a) is a strong standard to guarantee the privacy for algorithms that have been widely applied to modern machine learning (Abadi et al., 2016). It characterizes the privacy loss via statistical hypothesis testing, thus allowing the mathematically rigorous analysis of the privacy bounds. When multiple operations on the data are involvedand each intermediate step is revealed, composition theorems can be used to keep track of the privacy loss, which combines subadditively (Kairouz et al., 2015). However, because such results are required to be general, their associated privacy bounds are inevitably loose. In contrast, privacy amplification provides a privacy budget for a composition of mechanisms that is less that the budget of each individual operation, which strengthens the bound the more operations are concatenated. Classic examples of this feature are privacy amplification by subsampling (Chaudhuri and Mishra, 2006; Balle et al., 2018), by shuffling (Erlingsson et al., 2019) and by iteration (Feldman et al., 2018; Asoodeh et al., 2020). In this paper, we focus on the setting of privacy amplification by iteration, and extend the analysis via contraction coefficient proposed by Asoodeh et al. (2020) to prove results that apply to an algorithm commonly used in practice, in which the entire dataset is 1

We show a simple reduction which demonstrates the cryptographic hardness of learning a single periodic neuron over isotropic Gaussian distributions in the presence of noise. More precisely, our reduction shows that any polynomial-time algorithm (not necessarily gradient-based) for learning such functions under small noise implies a polynomial-time quantum algorithm for solving worst-case lattice problems, whose hardness form the foundation of lattice-based cryptography. Our core hard family of functions, which are well-approximated by one-layer neural networks, take the general form of a univariate periodic function applied to an affine projection of the data. These functions have appeared in previous seminal works which demonstrate their hardness against gradient-based (Shamir'18), and Statistical Query (SQ) algorithms (Song et al.'17). We show that if (polynomially) small noise is added to the labels, the intractability of learning these functions applies to all polynomial-time algorithms under the aforementioned cryptographic assumptions. Moreover, we demonstrate the necessity of noise in the hardness result by designing a polynomial-time algorithm for learning certain families of such functions under exponentially small adversarial noise. Our proposed algorithm is not a gradient-based or an SQ algorithm, but is rather based on the celebrated Lenstra-Lenstra-Lov\'asz (LLL) lattice basis reduction algorithm. Furthermore, in the absence of noise, this algorithm can be directly applied to solve CLWE detection (Bruna et al.'21) and phase retrieval with an optimal sample complexity of $d+1$ samples. In the former case, this improves upon the quadratic-in-$d$ sample complexity required in (Bruna et al.'21). In the latter case, this improves upon the state-of-the-art AMP-based algorithm, which requires approximately $1.128d$ samples (Barbier et al.'19).

As the analytic tools become more powerful, and more data are generated on a daily basis, the issue of data privacy arises. This leads to the study of the design of privacy-preserving machine learning algorithms. Given two objectives, namely, utility maximization and privacy-loss minimization, this work is based on two previously non-intersecting regimes -- Compressive Privacy and multi-kernel method. Compressive Privacy is a privacy framework that employs utility-preserving lossy-encoding scheme to protect the privacy of the data, while multi-kernel method is a kernel based machine learning regime that explores the idea of using multiple kernels for building better predictors. The compressive multi-kernel method proposed consists of two stages -- the compression stage and the multi-kernel stage. The compression stage follows the Compressive Privacy paradigm to provide the desired privacy protection. Each kernel matrix is compressed with a lossy projection matrix derived from the Discriminant Component Analysis (DCA). The multi-kernel stage uses the signal-to-noise ratio (SNR) score of each kernel to non-uniformly combine multiple compressive kernels. The proposed method is evaluated on two mobile-sensing datasets -- MHEALTH and HAR -- where activity recognition is defined as utility and person identification is defined as privacy. The results show that the compression regime is successful in privacy preservation as the privacy classification accuracies are almost at the random-guess level in all experiments. On the other hand, the novel SNR-based multi-kernel shows utility classification accuracy improvement upon the state-of-the-art in both datasets. These results indicate a promising direction for research in privacy-preserving machine learning.

In this paper, we address Novel Class Discovery (NCD), the task of unveiling new classes in a set of unlabeled samples given a labeled dataset with known classes. We exploit the peculiarities of NCD to build a new framework, named Neighborhood Contrastive Learning (NCL), to learn discriminative representations that are important to clustering performance. Our contribution is twofold. First, we find that a feature extractor trained on the labeled set generates representations in which a generic query sample and its neighbors are likely to share the same class. We exploit this observation to retrieve and aggregate pseudo-positive pairs with contrastive learning, thus encouraging the model to learn more discriminative representations. Second, we notice that most of the instances are easily discriminated by the network, contributing less to the contrastive loss. To overcome this issue, we propose to generate hard negatives by mixing labeled and unlabeled samples in the feature space. We experimentally demonstrate that these two ingredients significantly contribute to clustering performance and lead our model to outperform state-of-the-art methods by a large margin (e.g., clustering accuracy +13% on CIFAR-100 and +8% on ImageNet).