Spain

A central challenge to contemporary AI is to integrate learning and reasoning. The integration of learning and reasoning has been studied for decades already in the fields of statistical relational artificial intelligence and probabilistic programming. StarAI has focussed on unifying logic and probability, the two key frameworks for reasoning, and has extended this probabilistic logics machine learning principles.
I will argue that StarAI and Probabilistic Logics form an ideal basis for developing neuro-symbolic artificial intelligence techniques. Thus neuro-symbolic computation = StarAI + Neural Networks.
Many parallels will be drawn between these two fields and will be illustrated using the Deep Probabilistic Logic Programming language DeepProbLog.

ECAI 2020

From Statistical Relational to Neural Symbolic Artificial Intelligence 

ECAI2020 featured a full programme of technical papers, as well as many AI-related events of interest for researchers, students and all attendants who are interested on contemporary AI, including:




**An extensive programme of workshops, tutorials, and invited speakers**

**The Prestigious Applications of Intelligent Systems conference (PAIS)**

**The Frontiers in AI track sessions**

**The Lab To Market event**

**The Women in AI time**

**The EU Challenges forum**

… and many more!

ECAI 2020 featured a full programme of technical papers, as well as many AI-related events of interest for researchers, students and all attendants who are interested on contemporary AI.

technical paper

This tutorial aims to provide an initial foundation in the concepts underlying computational argumentation and its role and relevance in the context of Human-Centric AI. We will also discuss potential applications in this context.

Computational Argumentation in the context of Human-Agent Interaction

Scalable Deep Learning: How far is one billion neurons?

Practical Uses of Existential Rules in Knowledge Representation

Tutorial on Distributed Constraint Optimization for the Internet-of-Things (DCOP for IoT)

Knowledge graph embeddings are supervised learning models that learn vector representations of nodes and edges of labeled, directed multigraphs. We describe their design rationale, and explain why they are receiving growing attention within the burgeoning graph representation learning community. We highlight their limitations, open research directions, and real-world applicative scenarios. Besides a theoretical overview, we also provide a hands-on session, where we show how to use such models in practice. https://kge-tutorial-ecai2020.github.io/

Knowledge Graph Embeddings: From Theory to Practice

"This tutorial teaches the fundamental tools and techniques for running GPU-accelerated Python applications using CUDA® and the NUMBA compilers. Each participant will have access to a GPU workstation in the cloud on which to complete the many example programs used to demonstrate the concepts. Participants will work through dozens of hands-on coding exercises, and at the end of the training, implement a new workflow to accelerate an example program originally designed for CPUs, observing impressive performance gains.
On completion of this tutorial, participants will have additional resources to help them create new GPU-accelerated applications on their own.  Moreover, upon successful completion of the final assessment, participants will receive an NVIDIA Deep Learning Institute (DLI) certificate to recognize their subject matter competency and support professional career growth.
The content for this tutorial has been developed by NVIDIA’s Deep Learning Institute (https://www.nvidia.com/en-us/deep-learning-ai/education/)

The tutorial covers 3 main topics (note that if not all these topics can be covered in the time allocated for the workshop, participants can complete them through self-study thereafter):

1.	Introduction to CUDA Python with Numba: 
a.	Begin working with the Numba compiler and CUDA programming in Python.
b.	Use Numba decorators to GPU-accelerate numerical Python functions.
c.	Optimize host-to-device and device-to-host memory transfers.
2.	Create CUDA kernels in Python with Numba:
a.	Learn CUDA’s parallel thread hierarchy and how to extend parallel program possibilities.
b.	Launch massively parallel custom CUDA kernels on the GPU.
c.	Utilize CUDA atomic operations to avoid race conditions during parallel execution.
3.	Multi-dimensional grids and shared memory for CUDA Python with Numba:
a.	Learn support for GPU-accelerated Monte Carlo methods.
b.	Learn multidimensional grid creation and how to work in parallel on 2D matrices.
c.	Leverage on-device shared memory to promote memory coalescing while reshaping 2D matrices.

The tutorial is aimed at anyone wishing to accelerate Python applications. By making use of the Numba (just-in-time) compiler, CPU-based applications can be easily modified to execute on a GPU with possible performance gains. 
Prerequisites include: 

1.	Basic Python competency, including familiarity with variable types, loops, conditional statements, functions, and array manipulations. 
2.	NumPy competency, including the use of ndarrays and ufuncs. 
Note that no previous knowledge of CUDA programming is required.
Please note that prior to the start of the tutorial, all participants should have created an account on the NVIDIA developer site:  courses.nvidia.com/join 
"

Fundamentals of Accelerated Computing with CUDA Python

Mathematical Morphology and Logics: In propositional logics, considering the lattice of formulas, morphological operators will act on formulas (and on their models). In modal logics, dilation and erosion can define modal operators. Such examples will be described in several logics. Then we will show how they can be used to define revision operators satisfying the AGM postulates, merging operators, or explanatory relations.

Mathematical Morphology and Artificial Intelligence

Cognitive Logics: Mechanisms Predicting Human Inference Patterns

Since the early days of AI, Knowledge Representation and Reasoning (KR) has pursued the goal of capturing human knowledge in forms that can be stored in machines and used for automated inference, enabling computers to draw conclusions analogous to the ones we humans draw from our own knowledge. Although some goals of the field’s original research agenda proved elusive, KR has delivered solutions to many central AI problems. In this talk, we will discuss a few selected areas where KR has proved very successful, if not necessarily at replicating the whole spectrum of human intelligence, certainly at achieving more intelligent information systems. For example, explicit knowledge captured in ontologies is a powerful tool to access information on the Web, and for facilitating the management of data that may be unstructured, heterogeneous, and incomplete. When combined with rules, it can be leveraged for solving hard problems under incomplete information. Beyond these success stories, we will also touch on some current AI challenges where we believe KR can play a central role: can explicit knowledge and automated reasoning help us make modern AI more transparent, safer, and more trustworthy?

Frontiers in AI talk: Knowledge and Reasoning for Intelligent Systems

In a dynamic modal logic a 'lie that p' (where p is a some proposition) is an action, that is interpreted as a state transformer relative to the proposition p. These states are pointed Kripke models encoding the uncertainty of agents about their beliefs, and their transformation results in updated beliefs. Lies can be about factual propositions but also about the beliefs of other agents. Deception can be given meaning in terms of protocols consisting of sequences of such actions, in view of realizing an epistemic goal. Agents can have many different roles. Two speaker perspectives are: (obs) an outside observer who is lying to an agent that is modelled in the system, and (ag) an agent who is lying to another agent, and where both are modelled in the system. Three addressee perspectives are: the *credulous* agent who believes everything that it is told (even at the price of inconsistency), the *skeptical* agent who only believes what it is told if that is consistent with its current beliefs, and the *belief revising* agent who believes everything that it is told by consistently revising its current, possibly conflicting, beliefs. Then again there may be non-addressed agents perceiving lies and deception yet different from the addressee. Lying may be costly. Not only in terms of delayed response time in psychological experiments, on which we will not speak, but also in terms of computational complexity of performing certain tasks while lying or deceiving, or while taking into account that one might be lied to. We do not know of hard results in this area but there seem to be many interesting open questions. Results on computational complexity seem to be particularly relevant for AI. Issues are not merely the possibility of one lie among many truths coming out of the mouth of a single agent (we recall Ulam games), but also the presence of one unreliable agent in the presence of many trustworthy agents (typical in security protocol settings). How to detect lies or liars, and how costly is that? There is ample room for special case studies and benchmarks, such as lying in gossip protocols. Background: Hans van Ditmarsch, Petra Hendriks, Rineke Verbrugge: Editors' Review and Introduction: Lying in Logic, Language, and Cognition. topiCS 12(2): 466-484 (2020)

Downloads

Next from ECAI 2020

Computational Argumentation in the context of Human-Agent Interaction

Similar lecture

Relational Neural Machines

.css-70qvj9{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;}Downloads

Next from ECAI 2020

Computational Argumentation in the context of Human-Agent Interaction

Similar lecture

Relational Neural Machines

Downloads