New scholarships offer students from underrepresented backgrounds the chance of postgraduate study in AI and machine learning. Interdisciplinary artificial intelligence group DeepMind has renewed its philanthropic commitment to postgraduate students from underrepresented backgrounds who wish to study AI and machine learning at Imperial's Department of Computing. A gift to the Department from DeepMind will support six new postgraduate scholarships in AI and machine learning over the next few academic years. DeepMind's gift provides funding for four Master's-level scholarships and two PhD scholarships, all with a focus on artificial intelligence and machine learning. The scholarships are targeted at groups who are currently underrepresented in computing, particularly women and those from black and minority ethnic backgrounds.

GlobalData research has found the top big data influencers based on their performance and engagement online. Using research from GlobalData's Influencer platform, Verdict has named ten of the most influential people in big data on Twitter during Q2 2020. Evan Kirstel is a B2B thought leader with extensive experience across enterprises sales, alliances, and business development. He currently serves as chief digital officer and advisor of NYDLA.ORG, a remote, distance/digital learning and collaboration association. Kirstel highlights the challenges of data ingestion within the healthcare context, and also stated that enterprises are collecting massive amounts of data but do not how to leverage it effectively.

With artificial intelligence and analytics being the talk of the hour, there cannot be a better time to get started with these technologies. COVID pandemic outbreak has further increased the demand for data scientists thus learning data science skills, in the current situation, can present high employment chances. Till now, the field of data science and AI has been a preferred choice for postgraduate programs; however, the increasing demand for data professionals is making it imperative for students to start early. And that's where an undergraduate AI and data science course can help. Now that the results for Class XII board exams are out, this could be the perfect chance for the pass out students to build a career in the most demanding profession of the world.

Intrinsically motivated spontaneous exploration is a key enabler of autonomous lifelong learning in human children. It enables the discovery and acquisition of large repertoires of skills through self-generation, self-selection, self-ordering and self-experimentation of learning goals. We present an algorithmic approach called Intrinsically Motivated Goal Exploration Processes (IMGEP) to enable similar properties of autonomous or self-supervised learning in machines. The IMGEP algorithmic architecture relies on several principles: 1) self-generation of goals, generalized as fitness functions; 2) selection of goals based on intrinsic rewards; 3) exploration with incremental goal-parameterized policy search and exploitation of the gathered data with a batch learning algorithm; 4) systematic reuse of information acquired when targeting a goal for improving towards other goals. We present a particularly efficient form of IMGEP, called Modular Population-Based IMGEP, that uses a population-based policy and an object-centered modularity in goals and mutations. We provide several implementations of this architecture and demonstrate their ability to automatically generate a learning curriculum within several experimental setups including a real humanoid robot that can explore multiple spaces of goals with several hundred continuous dimensions. While no particular target goal is provided to the system, this curriculum allows the discovery of skills that act as stepping stone for learning more complex skills, e.g. nested tool use. We show that learning diverse spaces of goals with intrinsic motivations is more efficient for learning complex skills than only trying to directly learn these complex skills.

We describe a prototype of a new experimental GeoGebra command and tool Discover that analyzes geometric figures for salient patterns, properties, and theorems. This tool is a basic implementation of automated discovery in elementary planar geometry. The paper focuses on the mathematical background of the implementation, as well as methods to avoid combinatorial explosion when storing the interesting properties of a geometric figure.

Graphical models are powerful tools for modeling high-dimensional data, but learning graphical models in the presence of latent variables is well-known to be difficult. In this work we give new results for learning Restricted Boltzmann Machines, probably the most well-studied class of latent variable models. Our results are based on new connections to learning two-layer neural networks under $\ell_{\infty}$ bounded input; for both problems, we give nearly optimal results under the conjectured hardness of sparse parity with noise. Using the connection between RBMs and feedforward networks, we also initiate the theoretical study of $supervised~RBMs$ [Hinton, 2012], a version of neural-network learning that couples distributional assumptions induced from the underlying graphical model with the architecture of the unknown function class. We then give an algorithm for learning a natural class of supervised RBMs with better runtime than what is possible for its related class of networks without distributional assumptions.

We present Gaptron, a randomized first-order algorithm for online multiclass classification. In the full information setting we show expected mistake bounds with respect to the logistic loss, hinge loss, and the smooth hinge loss with constant regret, where the expectation is with respect to the learner's randomness. In the bandit classification setting we show that Gaptron is the first linear time algorithm with $O(K\sqrt{T})$ expected regret, where $K$ is the number of classes. Additionally, the expected mistake bound of Gaptron does not depend on the dimension of the feature vector, contrary to previous algorithms with $O(K\sqrt{T})$ regret in the bandit classification setting. We present a new proof technique that exploits the gap between the zero-one loss and surrogate losses rather than exploiting properties such as exp-concavity or mixability, which are traditionally used to prove logarithmic or constant regret bounds.

Control schemes that learn using measurement data collected online are increasingly promising for the control of complex and uncertain systems. However, in most approaches of this kind, learning is viewed as a side effect that passively improves control performance, e.g., by updating a model of the system dynamics. Determining how improvements in control performance due to learning can be actively exploited in the control synthesis is still an open research question. In this paper, we present AntLer, a design algorithm for learning-based control laws that anticipates learning, i.e., that takes the impact of future learning in uncertain dynamic settings explicitly into account. AntLer expresses system uncertainty using a non-parametric probabilistic model. Given a cost function that measures control performance, AntLer chooses the control parameters such that the expected cost of the closed-loop system is minimized approximately. We show that AntLer approximates an optimal solution arbitrarily accurately with probability one. Furthermore, we apply AntLer to a nonlinear system, which yields better results compared to the case where learning is not anticipated.

Low-rank matrix completion is a popular paradigm in machine learning, but little is known about the completion properties of non-random observation patterns. A fundamental open question in this direction is the following: given an observation pattern of a sufficiently generic (e.g. incoherent) $m \times n$ real matrix $X$ of rank $r$ with exactly $r(m+n-r)$ entries being observed, this number being the dimension of the space of real rank-$r$ $m \times n$ matrices, are there finitely many rank-$r$ completions? This is a challenging problem whose answer is known only for ranks $1$, $2$ and $\min\{m,n\}-1$. In this paper we study this problem by bringing tools from algebraic geometry. In particular, we exploit the fact that both the space of real rank-$r$ $m \times n$ matrices as well as the set of $r$-dimensional subspaces of $\mathbb{R}^m$, known as the Grassmannian, are algebraic varieties. Our approach is based on a novel formulation of matrix completion in terms of Pl{\"u}cker coordinates, the latter a traditionally powerful tool in computer vision and graphics and a classical notion in algebraic geometry. This formulation allows us to characterize a large class of minimal (i.e. of size $r(m+n-r)$) observation patterns for which a generic matrix admits finitely many rank-r completions. We conjecture that the converse is also true: any minimal pattern which is generically finitely completable must be of that type. As a consequence, we generalize results that have previously appeared and are being used in the literature, but lack a sufficient theoretical justification. We believe the Pl{\"u}cker-coordinate based link that we establish between low-rank matrices and the Grassmannian in the context of matrix completion to be of wider significance for matrix and subspace learning problems with incomplete data.

Former U.S. President Obama put forth the initiative'CSForAll' in order to prepare all students to learn computer science (CS) skills and be prepared for the digital economy. The'ForAll' portion of the title emphasizes the importance of inclusion in computing via the participation and creation of tools by and for diverse populations in order to "avoid the consequences of narrowly focused AI (computing and other) applications, including the risk of biases in developing algorithms, by taking advantage of a broader spectrum of experience, backgrounds, and opinions."10 Throughout this report, the Obama administration highlighted the number one priority, and challenge, of the field of CS: to equip the next generation with CS knowledge and skills equitably in preparation for the currency of the digital economy. An increase in government funding is part of the initiative for CSForAll. Of the $4 billion pledged in state funding, only $100 million is sent directly to the K–12 school system.17 The rest of the funding is set aside for research and initiatives involving policymakers to help expand CS opportunities. In just one year, the National Science Foundation (NSF) and Corporation for National and Community Service (CNCS) were called to make $135 million in CS funding available.17 The initiative also called for "expanding access to prior NSF supported programs and professional learning communities through their CS10k that led to the creation of more inclusive and accessible CS education curriculum including "Exploring CS and Advanced Placement (AP) CS Principles."