Computational Intelligence Methods for Bioinformatics and Biostatistics, 2015
In this work we propose a novel hybrid technique for overlapping community detection in biologica... more In this work we propose a novel hybrid technique for overlapping community detection in biological networks able to exploit both the available quantitative and the semantic information, that we call Semantically Enriched Fuzzy C-Means Spectral Modularity (SE-FSM) community detection method. We applied SE-FSM in analyzing Protein-protein interactions (PPIs) networks of HIV-1 infection and Leukemia in Homo sapiens. SE-FSM found significant overlapping biological communities. In particular, it found a strong relationship between HIV-1 and Leukemia as their communities share several significant pathways, and biological functions.
Proceedings of International Conference on Neural Networks (ICNN'97), 1997
Abstract The paper presents a hardware implementation of the neural gas (NGAS) algorithm. The NGA... more Abstract The paper presents a hardware implementation of the neural gas (NGAS) algorithm. The NGAS is based on vector quantization and is applied to very low bit-rate video compression. The algorithm exhibits interesting properties that can be exploited in an HW realization. The modular structure provides inherent parallelism and can therefore be regarded as an open architecture. The neuro-board interfaces to a PC through a standard ISA bus. The novelty of the proposed solution lies in providing a PC-based configurable ...
Proceedings of International Conference on Neural Networks (ICNN'96), 1996
Abstract A parallel implementation of unsupervised vector-quantization networks can reduce the hi... more Abstract A parallel implementation of unsupervised vector-quantization networks can reduce the high computational load of the training process. First, a plastic version of the neural gas algorithm is presented. Then, the paper describes how a toroidal mesh topology fits the neural model for a distributed implementation. The architecture adopted and the data-allocation strategy enhance the method's scaling properties and remarkable efficiency. Experimental results on a significant testbed (low bit-rate image compression) confirm the ...
ABSTRACT When compared to standard clustering, fuzzy clustering provides more flexible and powerf... more ABSTRACT When compared to standard clustering, fuzzy clustering provides more flexible and powerful data representation. Most fuzzy methods require setting some parameters, as is the case for the Graded Possibilistic c-Means clustering method, which has two parameters in addition to number of centroids. However, for this model selection task there is no well established criterion available. Building on our own previous work on fuzzy clustering similarity indexes, we introduce a technique to evaluate the stability of clusterings by using the fuzzy Jaccard index, and use this procedure to select the most suitable values of parameters. The experiments indicate that the procedure is effective for fuzzy and possibilistic clustering.
IEE Proceedings - Circuits, Devices and Systems, 1999
Abstract A CMOS circuit for sorting analogue current-mode quantities is presented. The highly mod... more Abstract A CMOS circuit for sorting analogue current-mode quantities is presented. The highly modular architecture integrates several elementary cells operating at the local level. The VLSI-oriented approach minimises wiring and silicon area, because very few devices are involved. The sorting process is completed in O (n) time. Simulations at the VLSI layout level prove the effectiveness of the approach in neural-network training applications
Abstract The Letter describes the implementation of a high-connectivity processing node by means ... more Abstract The Letter describes the implementation of a high-connectivity processing node by means of an embedded shared dual-port memory. The memory is accessed directly by two transputers in order to realise a virtual processor with a connectivity and a computing power that are twice those of a single transputer
Abstract The K-winner machine (KWM) model for supervised classification enhances vector quantisat... more Abstract The K-winner machine (KWM) model for supervised classification enhances vector quantisation by characterising classification outcomes with confidence levels. Each data-space location is assigned a specific local bound to the error probability. Structural simplicity makes the implementation compatible with circuitry for classical VQ, and features high speed and efficiency
The paper reconsiders the applicability of Vector Quantization (VQ) to image compression for low ... more The paper reconsiders the applicability of Vector Quantization (VQ) to image compression for low bit-rate image transmission. The proposed method overcomes the basic, structural drawbacks of VQ by a general multiple-interpolation mechanism. The major advantages of the described schema are an improved generalization performance and a notable reduction in coarseness. The overall approach can then be integrated with classical adaptive methods to derive a flexible and effective compression schema without affecting ...
Large-scale parallelism and analog computation are exploited to obtain a neural module, suitable ... more Large-scale parallelism and analog computation are exploited to obtain a neural module, suitable for both functioning and training, since appropriate signal lines are provided. The VQ encoder is self-contained and therefore can be embedded into any system, either analog or digital. It implements efficiently the vector matching operations, therefore it can be exploited in systems based on any vector quantization algorithm, with good throughput.
The paper describes a board-based hardware implementation of a neural algorithm performing vector... more The paper describes a board-based hardware implementation of a neural algorithm performing vector quantization for very low bit-rate video compression. The Neural Gas model has been chosen for its remarkable properties in terms of both consistency (quality of the quantization process) and easy implementation. The neuroboard interfaces to a PC through a standard ISA bus. The board supports both training (codevector adjustment) and run-time operation. The main advantages of the implemented solution lie In its simplicity ...
Computational Intelligence Methods for Bioinformatics and Biostatistics, 2015
In this work we propose a novel hybrid technique for overlapping community detection in biologica... more In this work we propose a novel hybrid technique for overlapping community detection in biological networks able to exploit both the available quantitative and the semantic information, that we call Semantically Enriched Fuzzy C-Means Spectral Modularity (SE-FSM) community detection method. We applied SE-FSM in analyzing Protein-protein interactions (PPIs) networks of HIV-1 infection and Leukemia in Homo sapiens. SE-FSM found significant overlapping biological communities. In particular, it found a strong relationship between HIV-1 and Leukemia as their communities share several significant pathways, and biological functions.
Proceedings of International Conference on Neural Networks (ICNN'97), 1997
Abstract The paper presents a hardware implementation of the neural gas (NGAS) algorithm. The NGA... more Abstract The paper presents a hardware implementation of the neural gas (NGAS) algorithm. The NGAS is based on vector quantization and is applied to very low bit-rate video compression. The algorithm exhibits interesting properties that can be exploited in an HW realization. The modular structure provides inherent parallelism and can therefore be regarded as an open architecture. The neuro-board interfaces to a PC through a standard ISA bus. The novelty of the proposed solution lies in providing a PC-based configurable ...
Proceedings of International Conference on Neural Networks (ICNN'96), 1996
Abstract A parallel implementation of unsupervised vector-quantization networks can reduce the hi... more Abstract A parallel implementation of unsupervised vector-quantization networks can reduce the high computational load of the training process. First, a plastic version of the neural gas algorithm is presented. Then, the paper describes how a toroidal mesh topology fits the neural model for a distributed implementation. The architecture adopted and the data-allocation strategy enhance the method's scaling properties and remarkable efficiency. Experimental results on a significant testbed (low bit-rate image compression) confirm the ...
ABSTRACT When compared to standard clustering, fuzzy clustering provides more flexible and powerf... more ABSTRACT When compared to standard clustering, fuzzy clustering provides more flexible and powerful data representation. Most fuzzy methods require setting some parameters, as is the case for the Graded Possibilistic c-Means clustering method, which has two parameters in addition to number of centroids. However, for this model selection task there is no well established criterion available. Building on our own previous work on fuzzy clustering similarity indexes, we introduce a technique to evaluate the stability of clusterings by using the fuzzy Jaccard index, and use this procedure to select the most suitable values of parameters. The experiments indicate that the procedure is effective for fuzzy and possibilistic clustering.
IEE Proceedings - Circuits, Devices and Systems, 1999
Abstract A CMOS circuit for sorting analogue current-mode quantities is presented. The highly mod... more Abstract A CMOS circuit for sorting analogue current-mode quantities is presented. The highly modular architecture integrates several elementary cells operating at the local level. The VLSI-oriented approach minimises wiring and silicon area, because very few devices are involved. The sorting process is completed in O (n) time. Simulations at the VLSI layout level prove the effectiveness of the approach in neural-network training applications
Abstract The Letter describes the implementation of a high-connectivity processing node by means ... more Abstract The Letter describes the implementation of a high-connectivity processing node by means of an embedded shared dual-port memory. The memory is accessed directly by two transputers in order to realise a virtual processor with a connectivity and a computing power that are twice those of a single transputer
Abstract The K-winner machine (KWM) model for supervised classification enhances vector quantisat... more Abstract The K-winner machine (KWM) model for supervised classification enhances vector quantisation by characterising classification outcomes with confidence levels. Each data-space location is assigned a specific local bound to the error probability. Structural simplicity makes the implementation compatible with circuitry for classical VQ, and features high speed and efficiency
The paper reconsiders the applicability of Vector Quantization (VQ) to image compression for low ... more The paper reconsiders the applicability of Vector Quantization (VQ) to image compression for low bit-rate image transmission. The proposed method overcomes the basic, structural drawbacks of VQ by a general multiple-interpolation mechanism. The major advantages of the described schema are an improved generalization performance and a notable reduction in coarseness. The overall approach can then be integrated with classical adaptive methods to derive a flexible and effective compression schema without affecting ...
Large-scale parallelism and analog computation are exploited to obtain a neural module, suitable ... more Large-scale parallelism and analog computation are exploited to obtain a neural module, suitable for both functioning and training, since appropriate signal lines are provided. The VQ encoder is self-contained and therefore can be embedded into any system, either analog or digital. It implements efficiently the vector matching operations, therefore it can be exploited in systems based on any vector quantization algorithm, with good throughput.
The paper describes a board-based hardware implementation of a neural algorithm performing vector... more The paper describes a board-based hardware implementation of a neural algorithm performing vector quantization for very low bit-rate video compression. The Neural Gas model has been chosen for its remarkable properties in terms of both consistency (quality of the quantization process) and easy implementation. The neuroboard interfaces to a PC through a standard ISA bus. The board supports both training (codevector adjustment) and run-time operation. The main advantages of the implemented solution lie In its simplicity ...
Uploads