1,720,967 research outputs found
Using Contextual Information In Weakly Supervised Learning: Toward the integration of contextual and deep learningapproaches, to address weakly supervised tasks
No full textCome l'attento lettore avrà dedotto dal titolo, questa tesi pone alcune basi empiriche, assieme ad altrettante considerazioni teoriche, verso la definizione di una metodologia finalizzata a migliorare task di weakly supervised learning. La metodologia genera supervisione addizionale sfruttando l'informazione contestuale proveniente dal confronto delle osservazioni in un dataset sotto molteplici ipotesi di etichettatura.
Il materiale di ricerca presentato, ruota principalmente attorno a due algoritmi. Nella prima parte, l'attenzione è rivolta a Graph Transduction Games (GTG), un algoritmo di label propagation basato su nozioni di Teoria dei Giochi. In particolare, questo documento descrive le interazioni sperimentate con GTG e dei deep feature extractor, per affrontare problemi di semi-supervised, domain adaptation e deep metric learning. La seconda parte è incentrata su Relaxation Labeling (ReLab), una famiglia di processi utilizzata per label disambiguation, fortemente connessa a GTG, sebbene sia motivata da un differente contesto teorico. Questo documento alcuni concetti preliminari di teoria e degli esperimenti pensati per investigare future applicazioni di ReLab nel contesto di semi-supervised semantic segmentation.
Il lavoro presentato di seguito può essere pensato come un punto iniziale per costituire una teoria di contextual weakly supervised learning
On generalized KKT points of the Motzkin-Straus program
No full textIn 1965, Motzkin and Straus established a profound connection between the clique number of a graph and the global maxima of a quadratic program defined on the standard simplex. Since then, a line of active and intensive research has been yielding heuristics and bounds concerning the maximum clique problem, thanks to the discoveries pertaining local/global solutions of the Motzkin-Straus program. However, the Karush-Kuhn-Tucker (KKT) points thereof have received little to no attention in the literature. In this work, a parameterized version of the Motzkin-Straus program is discussed, and some results about its KKT points are obtained. What emerges is a connection between a generalized notion of KKT point and some regular structures contained in the graph
On generalized KKT points for the Motzkin-Straus program
No full textIn 1965, T. S. Motzkin and E. G. Straus established an elegant connection between the clique number of a graph and the global maxima of a quadratic program defined on the standard simplex. Over the years, this seminal finding has inspired a number of studies aimed at characterizing the properties of the (local and global) solutions of the Motzkin-Straus program. The result has also been generalized in various ways and has served as the basis for establishing new bounds on the clique number and developing powerful clique-finding heuristics. Despite the extensive work done on the subject, apart from a few exceptions, the existing literature pays little or no attention to the Karush-Kuhn-Tucker (KKT) points of the program. In the conviction that these points might reveal interesting structural properties of the graph underlying the program, this paper tries to fill in the gap. In particular, we study the generalized KKT points of a parameterized version of the Motzkin-Straus program, which are defined via a relaxation of the usual first-order optimality conditions, and we present a number of results that shed light on the symmetries and regularities of certain substructures associated with the underlying graph. These combinatorial structures are further analyzed using barycentric coordinates, thereby providing a link to a related quadratic program that encodes local structural properties of the graph. This turns out to be particularly useful in the study of the generalized KKT points associated with a certain class of graphs that generalize the notion of a star graph. Finally, we discuss the associations between the generalized KKT points of the Motzkin-Straus program and the so-called replicator dynamics, thereby offering an alternative, dynamical-system perspective on the results presented in the paper.27 pages, 3 figure
A Computer Vision System for Monitoring Ice-Cream Freezers
No full textIn this paper, we describe a computer vision system aimed at monitoring the evolution of the content of a commercial ice-cream freezer. In particular, the system is able to detect the volume occupied by ice-creams in a basket and to track ice-cream sales. To this end, three modules have been developed performing the detection of the baskets and the products inside them, along with the tracking of the interactions with the freezer to take/drop products. The system comprises four cameras connected to an embedded mini-computer able to communicate with a telemetry system that sends information about the freezer context. Our proposed methods achieve promising results for the basket detection and the product tracking (accuracy around 70–80%) and good results in the volume estimation
Exploiting Context in Handwriting Recognition Using Trainable Relaxation Labeling
No full textHandwriting Text Recognition (HTR) is a fast-moving research topic in computer vision and machine learning domains. Many models have been introduced over the years, one of the most well-established ones being the Convolutional Recurrent Neural Network (CRNN), which combines convolutional feature extraction with recurrent processing of the visual embeddings. Such a model, however, presents some limitations such as a limited capability to account for contextual information. To counter this problem, we propose a new learning module built on top of the convolutional part of a classical CRNN model, derived from the relaxation labeling processes, which is able to exploit the global context reducing the local ambiguities and increasing the global consistency of the prediction. Experiments performed on three well-known handwritten recognition datasets demonstrate that the relaxation labeling procedures improve the overall transcription accuracy at both character and word levels
On the Interplay between Strong Regularity and Graph Densification
Full text linkIn this paper we analyze the practical implications of Szemerédi’s regularity lemma in the preservation of metric information contained in large graphs. To this end, we present a heuristic algorithm to find regular partitions. Our experiments show that this method is quite robust to the natural sparsification of proximity graphs. In addition, this robustness can be enforced by graph densification
The Group Loss++: A deeper look into group loss for deep metric learning
Full text linkDeep metric learning has yielded impressive results in tasks such as clustering and image retrieval by leveraging neural networks to obtain highly discriminative feature embeddings, which can be used to group samples into different classes. Much research has been devoted to the design of smart loss functions or data mining strategies for training such networks. Most methods consider only pairs or triplets of samples within a mini-batch to compute the loss function, which is commonly based on the distance between embeddings. We propose Group Loss, a loss function based on a differentiable label-propagation method that enforces embedding similarity across all samples of a group while promoting, at the same time, low-density regions amongst data points belonging to different groups. Guided by the smoothness assumption that '`similar objects should belong to the same group'', the proposed loss trains the neural network for a classification task, enforcing a consistent labelling amongst samples within a class. We design a set of inference strategies tailored towards our algorithm, named Group Loss++ that further improve the results of our model. We show state-of-the-art results on clustering and image retrieval on four retrieval datasets, and present competitive results on two person re-identification datasets, providing a unified framework for retrieval and re-identification
Transductive Label Augmentation for Improved Deep Network Learning
No full textA major impediment to the application of deep learning to real-world problems is the scarcity of labeled data. Small training sets are in fact of no use to deep networks as, due to the large number of trainable parameters, they will very likely be subject to overfitting phenomena. On the other hand, the increment of the training set size through further manual or semi-automatic labellings can be costly, if not possible at times. Thus, the standard techniques to address this issue are transfer learning and data augmentation, which consists of applying some sort of 'transformation' to existing labeled instances to let the training set grow in size. Although this approach works well in applications such as image classification, where it is relatively simple to design suitable transformation operators, it is not obvious how to apply it in more structured scenarios. Motivated by the observation that in virtually all application domains it is easy to obtain unlabeled data, in this paper we take a different perspective and propose a label augmentation approach. We start from a small, curated labeled dataset and let the labels propagate through a larger set of unlabeled data using graph transduction techniques. This allows us to naturally use (second-order) similarity information which resides in the data, a source of information which is typically neglected by standard augmentation techniques. In particular, we show that by using known game theoretic transductive processes we can create larger and accurate enough labeled datasets which use results in better trained neural networks. Preliminary experiments are reported which demonstrate a consistent improvement over standard image classification datasets
Going Beyond Counting First Authors in Author Co-citation Analysis
Full text linkThe present study examines one of the fundamental aspects of author co-citation analysis (ACA) - the way co-citation
counts are defined. Co-citation counting provides the data on which all subsequent statistical analyses and mappings
are based, and we compare ACA results based on two different types of co-citation counting - the traditional type that
only counts the first one among a cited work's authors on the one hand and a non-traditional type that takes into
account the first 5 authors of a cited work on the other hand. Results indicate that the picture produced through this non-traditional author co-citation counting contains more coherent author groups and is therefore considerably clearer. However, this picture represents fewer specialties in the research field being studied than that produced through the traditional first-author co-citation counting when the same number of top-ranked authors is selected and analyzed. Reasons for these effects are discussed
- …
