Machine Perception

Analysis of musical styles is a complex cognitive task normally performed by music fans and critics, and due to the multi-dimensional nature of music data can be considered a challenging task for computing machines. Here we propose an automatic quantitative method that can analyze similarities between the sound of popular music albums in an unsupervised fashion. The method works by first converting the music samples into two-dimensional spectrograms, and then extracting a large set of 2883 2D numerical content descriptors from the raw spectrograms as well as 2D transforms and compound transforms of the spectrograms. The similarity between each pair of samples is computed using a variation of the Weighted K-Nearest Neighbor scheme, and a phylogeny is then used to visualize the differences between the albums. Experimental results show that the method was able to automatically organize the albums of The Beatles by their chronological order, and also unsupervisely arranged albums of musicians such as U2, Queen, ABBA, and Tears for Fears in a fashion that is largely in agreement with their chronological order and musical styles.

Convolutional neural networks can successfully perform many computer vision tasks on images. For visualization, how do CNNs perform when applied to graphical perception tasks? We investigate this question by reproducing Cleveland and McGill’s seminal 1984 experiments, which measured human perception efficiency of different visual encodings and defined elementary perceptual tasks for visualization. We measure the graphical perceptual capabilities of four network architectures on five different visualization tasks and compare to existing and new human performance baselines. While under limited circumstances CNNs are able to meet or outperform human task performance, we find that CNNs are not currently a good model for human graphical perception. We present the results of these experiments to foster the understanding of how CNNs succeed and fail when applied to data visualizations.

Neural networks and symbolic systems are two different approaches to model cognitive functions. Both methods have certain strengths but also suffer from a number of weak points, which are however disparate. By combining characteristics of neural and symbolic processing, these weaknesses could be overcome. This paper introduces a new information processing principle based on so-called neuro-symbolic networks, which incorporates these two approaches. The utility of the suggested principle is outlined for applications in machine perception.

Almtract-An algorithm for the perception of edge information by a vision machine is presented in this paper. In this approach, the sensored visual data are transformed by the process of cognition into generalized manifestations called percepts and concepts. Machine cognition is hypothesized as the transformation of sensory information between the information domain and comprehension domain. Percepts are the mental impressions of the sensory data which arise within the information domain. On the other hand, concepts are meanings which are attached to these percepts in the comprehension domain. The mechanisms for transforming raw data into percepts and then attaching conceptual meanings to them may be described by cognitive mapping functions. In this paper, a fuzzy set theoretic approach is used to emulate the cognitive mapping processes that occur within the information domain. More specifically, this algorithm employs fuzzy mapping functions to transform sensory gray-level images into various properties associated with a perceived edge. This new notion interprets an edge as a field with distributed membership values that may be observed at various levels of perception.

We propose that some aspects of task based learning in robotics can be approached using nativist and constructivist views on human sensorimotor development as a metaphor. We use findings in developmental psychology, neurophysiology, and machine perception to guide a robotic learning system's level of representation both for actions and for percepts. Visually driven grasping is chosen as the experimental task since it has general applicability and it has been extensively researched from several perspectives. An implementation of a robotic system with a dexterous three fingered hand, compliant instrumented wrist, arm and vision is used to test these ideas. Several sensorimotor primitives (vision segmentation and manipulatory reflexes) are implemented in this system and may be thought of as the "innate" perceptual and motor abilities of the system.

Artificial Intelligence (AI) is now widely used across a wide variety of fields. Yet, there are so far very few reported in-depth experimental studies utilising artificial intelligence in modelling cognitively of how the brain or mind processes aesthetics. In this paper, we document our study of aesthetical judgements via Artificial Neural Network (ANN). A technology inspired from neuronal processes of the brain is here enabled through training to derive meanings from symbolic designs. The insights gained from our cognitive neurodynamic modelling of the processes are presented. To the best of our knowledge, ours is probably the first prototypical artificial judge of aesthetics ever created: one capable of assessing designs in a human-like fashion.

Pengenalan objek (object recognition) merupakan salah satu image processing untuk mengenali objek-objek yang akan dikenali untuk diolah lebih lanjut agar mendapatkan suatu data informasi. Dengan adanya proses pengenalan objek tersebut, manusia akan di permudah dalam mengenali bentuk dari objek seperti pemetaan geografis, penggunaan sensor pada benda untuk sistem keamanan, pembacaan citra hasil scan medis, dan lain-lain.Penngenalan objek (object recognition) merupakan salah satu image processing untuk mengenali objek-objek yang akan dikenali untuk diolah lebih lanjut agar mendapatkan suatu data informasi. Terdapat beberapa metode deskripsi fitur (Feature Descriptor) untuk mendeteksi objek, seperti SIFT (Scale Invariant Feature Transformation), SURF (Speed Up Robust Feature), Color based object detection, HOG (Histograms of Oriented Gradients), Viola Jones, Optical Flow, dan lain-lain. Dalam penelitian ini, metode fitur ekstraksi yang digunakan adalah HOG (Histograms of Oriented Gradients) dan menggunakan SVM (Support Vector Machine) sebagai klasifikasi fitur (feature classifier). HOG digunakan dalam penelitian ini dikarenakan HOG merupakan Feature Descriptor yang mengambil tepi atau struktur gradient yang terkarakteristik dari bentuk lokal atau arah tepi dan dengan distribusi intensitas gradient lokal yang baik. Dalam penerapan objek deteksi terdapat beberapa penerapan sistem seperti di bidang keamanan, pengembangan Artificial Intelligent, pendataan secara otomatis (automatic data collect), dan lain-lain.
Adapun masalah yang menjadi permasalahan dalam penelitian tugas akhir ini adalah. Minimnya pengolahan informasi dari pendeteksian objek untuk pengembangan modul sistem cerdas pada Robot NAO. Dalam mendeteksi objek terdapat beberapa aspek yang mempengaruhi akurasi dan pengenalan objek, seperti intensitas pencahayaan yang minimum dan  jumlah data training.
Tujuan dari penelitian tugas akhir ini adalah mengembangkan modul sistem cerdas pada Robot NAO dengan membangun aplikasi yang dapat mendeteksi objek di sekitar. Melakukan pengujian keakurasian deteksi terhadap algoritma Feature Descriptor Histogram of Oriented Gradient yang digunakan dalam membangun aplikasi eteksi objek untuk modul sistem cerdas pada Robot NAO.
Hasil dari Perancangan Aplikasi Deteksi Objek menggunakan algoritma Histogram of Oriented Gradient ini dapat digunakan untuk penambahan modul sistem cerdas pada Robot NAO di karenakan dari hasil pengujian menghasilkan pengujian tes model menunjukkan tingkat keberhasilan mencapai 99,10% dalam mengenali image tes. Pada pengujian image tes positif, jumlah penggunaan data training sangat mempengaruhi dalam proses pendeteksian objek menjadi lebih akurat. Pada pengujian image gabungan (tes image positif dan negatif) tingkat keberhasilannya naik 98.23% menjadi 99,10% .

Keyword : Image Processing, Computer Vision, Machine Perception, Feature Descriptor Algorithm, Histogram of Oriented Gradient, Support Vector Machine, Robot NAO, Artificial Intelegence For Robot.