Greece

Neural Architecture Search is becoming an increasingly popular research field and method to design deep learning architectures. Most research focuses on searching for small blocks of deep learning operations, or micro-search. This method yields satisfactory results but demands prior knowledge of the macro architecture&#39;s structure. Generally, methods that do not utilize macro structure knowledge perform worse but are able to be applied to datasets of completely new domains. In this paper, we propose a macro NAS methodology which utilizes concepts of Regularized Evolution and Macro Neural Architecture Search (DeepNEAT), and apply it to the Fashion-MNIST dataset. By utilizing our method, we are able to produce networks that outperform other macro NAS methods on the dataset, when the same post-search inference methods are used. Furthermore, we are able to achieve 94.46% test accuracy, while requiring considerably less epochs to fully train our network.

AIAI 2020

Regularized Evolution for Macro Neural Architecture Search

fashion-mnist

neural architecture search

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

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.

Neural Architecture Search is becoming an increasingly popular research field and method to design deep learning architectures. Most research focuses on searching for small blocks of deep learning operations, or micro-search. This method yields satisfactory results but demands prior knowledge of the macro architecture's structure. Generally, methods that do not utilize macro structure knowledge perform worse but are able to be applied to datasets of completely new domains. In this paper, we propose a macro NAS methodology which utilizes concepts of Regularized Evolution and Macro Neural Architecture Search (DeepNEAT), and apply it to the Fashion-MNIST dataset. By utilizing our method, we are able to produce networks that outperform other macro NAS methods on the dataset, when the same post-search inference methods are used. Furthermore, we are able to achieve 94.46% test accuracy, while requiring considerably less epochs to fully train our network.

technical paper

AIAI 2020 (16th International Conference on Artificial Intelligence Applications and Innovations) should be held in Halkidiki, Greece. Due to current Covid-19 situation the conference will be held virtually. Underline will host a repository of video talks and preserve them for the future. Videos will be released on Underline one day after the conference.

The recent advancements in the areas of computer vision and deep learning with the development of convolutional neural networks and the profusion of highly accurate general purpose pre-trained models, create new opportunities for the interaction of humans with systems and facilitate the development of advanced features for all types of platforms and applications. Research, consumer and industrial applications increasingly integrate deep learning frameworks into their operational flow, and thanks to the availability of high performance hardware (Computer Boards, GPUs, TPUs) also for individual consumers and home use, this functionality has been moved closer to the end-users, at the edge of the network. In this work, we exploit the aforementioned approaches and tools for the development of an edge-native platform for exergames, which includes innovative gameplay and features for the users. A prototype game was created using the platform that was deployed in the real-world scenario of a rehabilitation center. The proposed approach provides advanced user experience based on the automated, real-time pose and gesture detection, and in parallel maintains a low cost to enable wide adoption in multiple applications across domains and usage scenarios.

Introducing an edge-native deep learning platform for exergames

The increase of the adoption of IoT devices and the contemporary problem of food production have given rise to numerous applications of IoT in agriculture. These applications typically comprise a set of sensors that are installed in open fields and measure metrics, such as temperature or humidity, which are used for irrigation control systems. Though useful, most contemporary systems have high installation and maintenance costs, and they do not offer automated control or, if they do, they are usually not interpretable, and thus cannot be trusted for such critical applications. In this work, we design Vital, a system that incorporates a set of low-cost sensors, a robust data store, and most importantly an explainable AI decision support system. Our system outputs a fuzzy rule-base, which is interpretable and allows fully automating the irrigation of the fields. Upon evaluating Vital in two pilot cases, we conclude that it can be effective for monitoring open-field installations.

Versatile Internet of Things for Agriculture: An eXplainable AI Approach

In this presentation, we investigate a special case of an open problem that is related to the exchange principle, a property of fuzzy implications. We focus on the cases of (S,N)- implications and the preservation of the exchange principle via lattice operations. We present and prove some sufficient conditions such that the exchange principle is preserved under the join and meet operations if we use (S,N)-implications.

Preservation of the Exchange Principle via Lattice Operations on (S,N)-Implications

In fuzzy hypothesis testing we use
fuzzy test statistics produced by fuzzy estimators and fuzzy critical values.
In this paper we use the non-asymptotic fuzzy estimators in fuzzy hypothesis testing. These are triangular shaped fuzzy numbers that generalize the fuzzy estimators based on confidence intervals in such a way that eliminates discontinuities and ensures compact support.
Our approach is particularly useful in critical situations, where
subtle fuzzy comparisons between almost equal statistical quantities have to be made. In such cases the hypotheses tests that use non-asymptotic fuzzy estimators give better results than the previous approaches, since they give us the possibility of partial rejection or not of the null hypothesis.

Hypotheses tests using non-asymptotic fuzzy estimators and fuzzy critical values

In this presentation, we introduce a fusion framework for image tampering localization, that moves towards overcoming the limitation of available tools by allowing a synergistic analysis and multi-perspective refinement of the final forensic report. The framework is designed to combine multiple state-of-the-art techniques by exploiting their complementarities so as to produce a single refined tampering localization output map. Extensive evaluation experiments of state-of-the-art methods on diverse datasets have resulted in a modular framework design where candidate methods go through a multi-criterion selection process to become part of the framework. Currently, this includes a set of five passive tampering localization methods for splicing localization on JPEG images. Our experimental findings on two different benchmark datasets showcase that the fused output achieves high performance and advanced interpretability by managing to leverage the correctly localized outputs of individual methods, and even detecting cases that were missed by all individual methods.

Knowledge-based fusion for image tampering localization

We present a new model that performs arbitrary scale, photo-realistic Super-Resolution without compromising on performance, called Meta-SRGAN. Meta-SRGAN is also extremely lightweight and we show how these qualities aid in several applications of Meta-SRGAN, including its use for COVID-19.

Arbitrary Scale Super-Resolution for Brain MRI Images

The presentation talk about Transfer learning using Convolutional Neural Network Architectures for Brain tumors classification. First, I'll talk about why we are interested in automatic brain tumors classification Second, I’ll describe the nine pre-trained networks architectures and transfer learning setting .Then, I’ll show the dataset used, training parameters and how we calculate the classification accuracy. Next , I ‘ll also show our results which outperforms the results of the related works. finally, I’ll give you a brief conclusion and future works.

Transfer Learning using Convolutional Neural Network Architectures for Brain Tumor Classification from MRI Images

The study of Self-organizing lists deals with the problem of lowering the average-case asymptotic cost of a list data structure receiving query accesses in Non-stationary Environments (NSEs) with the so-called “locality of reference” property. The de facto schemes for Adaptive lists in such Environments are the Move To Front (MTF) and Transposition (TR) rules. However, significant drawbacks exist in the asymptotic accuracy and speed of list re-organization for the MTF and TR rules. This paper improves on these schemes using the design of an Adaptive list data structure as a hierarchical data “sub”-structure. In this framework, we employ a hierarchical Singly-Linked-Lists on Singly- Linked-Lists (SLLs-on-SLLs) design, which divides the list data structure into an outer and inner list context. The inner-list context is itself a SLLs containing sub-elements of the list, while the outer-list context contains these sublist partitions as its primitive elements. The elements belong- ing to a particular sublist partition are determined using reinforcement learning schemes from the theory of Learning Automata. In this pa- per, we show that the Transitivity Pursuit-Enhanced Object Migration Automata (TPEOMA) can be used in conjunction with the hierarchical SLLs-on-SLLs as the dependence capturing mechanism to learn the probabilistic distribution of the elements in the Environment. The idea of Transitivity builds on the Pursuit concept that injects a noise filter into the EOMA to filter divergent queries from the Environment, thereby increasing the likelihood of training the Automaton to approximate the “true” distribution of the Environment. By taking advantage of the Transitivity phenomenon based on the statistical distribution of the queried elements, we can infer “dependent” query pairs from non-accessed elements in the transitivity relation. The TPEOMA-enhanced hierarchical SLLs-on-SLLs schemes results in superior performances to the MTF and TR schemes as well as to the EOMA-enhanced hierarchical SLLs- on-SLLs schemes in NSEs. However, the results are observed to have superior performances to the PEOMA-enhanced hierarchical schemes in Environments with a Periodic non-stationary distribution but were inferior in Markovian Switching Environments.

Optimizing Self-Organizing Lists-on-Lists using Transitivity and Pursuit-Enhanced Object Partitioning

Predicting the activity of chemical compounds against
cancer is a crucial task. Active chemical compounds against cancer
help pharmaceutical drugs producers in the conception of anticancer
medicines. Still the innate way of representing chemical compounds is
by graphs, the machine learning algorithms can not handle directly the
anticancer activity prediction problems. In this presentation, we investigate the representation power of Node2vec for embedding learning over
graphs, by comparing it to the theoretical framework Graph Isomorphism Network (GIN). We prove that GIN is a deep generalization of
Node2vec. We then exert the two models Node2vec and GIN to extract
regular representations from chemical compounds and make predictions
about their activity against lung and ovarian cancer

Graph Neural Networks to Advance Anticancer Drug Design

Brain-inspired Hyperdimensional (HD) computing is an emerging technique for cognitive tasks in the field of low-power design. As an energy-efficient and fast learning computational paradigm, HD computing has shown great success in many real-world applications. However, an HD model incrementally trained on multiple tasks suffers from the negative impacts of catastrophic forgetting. The model forgets the knowledge learned from previous tasks and only focuses on the current one. To the best of our knowledge, no study has been conducted to investigate the feasibility of applying multi-task learning to HD computing. In this paper, we propose Task-Projected Hyperdimensional Computing (TP-HDC) to make the HD model simultaneously support multiple tasks by exploiting the redundant dimensionality in the hyperspace. To mitigate the interferences between different tasks, we project each task into a separate subspace for learning. Compared with the baseline method, our approach efficiently utilizes the unused capacity in the hyperspace and shows a 12.8% improvement in averaged accuracy with negligible memory overhead.

Regularized Evolution for Macro Neural Architecture Search

Downloads

Next from AIAI 2020

Introducing an edge-native deep learning platform for exergames

Similar lecture

Task-Aware Performance Prediction for Efficient Architecture Search

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES

Regularized Evolution for Macro Neural Architecture Search

.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 AIAI 2020

Introducing an edge-native deep learning platform for exergames

Similar lecture

Task-Aware Performance Prediction for Efficient Architecture Search

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES

Downloads