<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Hong Jun Chew on Medium]]></title>
        <description><![CDATA[Stories by Hong Jun Chew on Medium]]></description>
        <link>https://medium.com/@hjchew96?source=rss-7956c6ed626d------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*8AkPD-Q-zFdA5_7C.jpg</url>
            <title>Stories by Hong Jun Chew on Medium</title>
            <link>https://medium.com/@hjchew96?source=rss-7956c6ed626d------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 10:55:23 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@hjchew96/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Network Science [5]]]></title>
            <link>https://medium.com/@hjchew96/network-science-5-e0e31ab92e32?source=rss-7956c6ed626d------2</link>
            <guid isPermaLink="false">https://medium.com/p/e0e31ab92e32</guid>
            <dc:creator><![CDATA[Hong Jun Chew]]></dc:creator>
            <pubDate>Wed, 26 Feb 2020 07:19:46 GMT</pubDate>
            <atom:updated>2020-02-26T07:19:46.302Z</atom:updated>
            <content:encoded><![CDATA[<p>Network Centrality</p><p>General Measures of Networks Learnt</p><ol><li>Average Degree &lt;k&gt;</li><li>Degree Distribution P(k)</li><li>Diameter and Distance</li><li>Clustering Coefficient, Cᵢ</li></ol><p>Centrality Measures</p><ol><li>Degree Centrality → Normalized Degree Centrality</li><li>Eigenvector Centrality</li><li>Betweenness Centrality</li><li>Closeness Centrality</li></ol><ul><li>Degree Centrality takes values each node by its degree<br>Normalized centrality just takes that value divided by (N-1)<br>Freeman’s formula could be used to check the distribution of the degrees<br>↑ High Centrality: Hubs; <br>↓Low Centrality: Distributed</li><li>Eigenvector measures importance via adjacent vectors</li><li>Betweenness Centrality, measures which nodes do you have to go though<br>normalized , divided by [(n-1)(n-2)/2]</li><li>Closeness, “length of the shortest path between vertex and all other vertexes”, divided by N-1, take the inverse</li></ul><p>In conclusion, Degree measures the number of connections between nodes, whereas closeness measures the shortest path to all other nodes.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e0e31ab92e32" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[NetCentric Computing [1]]]></title>
            <link>https://medium.com/@hjchew96/netcentric-computing-1-dbf7cd5626c0?source=rss-7956c6ed626d------2</link>
            <guid isPermaLink="false">https://medium.com/p/dbf7cd5626c0</guid>
            <dc:creator><![CDATA[Hong Jun Chew]]></dc:creator>
            <pubDate>Mon, 17 Feb 2020 05:32:28 GMT</pubDate>
            <atom:updated>2020-02-17T05:32:28.644Z</atom:updated>
            <content:encoded><![CDATA[<ol><li><strong>Layered Network Architecture</strong><br>1.1. Motivation and Benefits<br>1.2. OSI 7-Layer Model<br>1.3. Internet 5-Layer Implementation</li><li>Physical Resilience<br>2.1 Link Failure probability<br>2.2 Network Resilience Calculation</li></ol><p>A network comprises of Hosts, Routers, Links, Applications, Protocols, hardware and software</p><p>The layering of the web<br>1. Client Browser <strong>requests </strong>a webpage from sever.<br>2. Sever <strong>determine </strong>access control.<br>3. <strong>Sets up </strong>a reliable transfer from the server to the client.<br>4. Physical <strong>transfer</strong> of bits from the server to the client.</p><p>For 3 to happen:</p><p>The sever and the client must have the same <strong>protocol:<br></strong>this can include format, the order of message sent and receiving and actions to be taken.</p><p>The benefits of having multiple layers in the web framework are to have clear segregation (Decomposition) of roles between each layer, allowing for easy modification and addition of new layers.</p><h3>OSI 7-Layer Model</h3><p>Benefits:</p><ol><li><strong>Functional Decomposition</strong>, weakly interdependent between layers</li><li><strong>Encapsulation</strong>, each layer has it’s own role</li><li><strong>Peering</strong>, each layer only communicates with adjacent layers</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/781/1*iMkzV-eYGsVBzX71el_UQQ.png" /><figcaption><a href="https://www.geeksforgeeks.org/layers-of-osi-model/">https://www.geeksforgeeks.org/layers-of-osi-model/</a></figcaption></figure><p>At the Network layer, Datalink layer <strong>headers will be added </strong>to each layer.<br>and at the physical layer, these data will be <strong>converted into physical bits</strong>.<br>Network Layer defines how to fragment a packet into smaller packets for different media</p><p>Layer N of the sever will interact with another layer N with the client. The corresponding layer is known as <strong>Peer Processes.<br></strong>This convention is also known as <strong>Layer-n protocol</strong>, and these layers communicate via <strong>Protocol Data Units (PDUs).<br></strong>This communication between Peer Processes is purely virtual, and the upper layers (n+1) get information by invoking the services (methods) on the lower layer (n). Services available are found at Service Acess Points.</p><p>Communication <strong>Downwards </strong>from upper layers (n) to the lower layers (n-1) is called <strong>Service Data Units</strong> (SDU) and is encapsulated in PDUs.</p><h3>TCP/IP Model: 5 Layers</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/666/1*IgwwHbx2xQoYoa-Wex1D3g.png" /><figcaption><a href="https://www.geeksforgeeks.org/tcp-ip-model/">https://www.geeksforgeeks.org/tcp-ip-model/</a></figcaption></figure><p>Unlike OSI, TCP/IP has only 5 separate layers</p><ol><li>Application Layer supports application via FTP, SMTP, and HTTP</li><li>Transport Layer, host to host data transfer via TCP, UDP</li><li>Internet Layer, routing of datagrams from source to destination, via IP, routing protocols</li><li>Network Access Layer, data transfer between network elements via PPP, Ethernet</li><li>Physical, bits on the wire</li></ol><p><strong>PHYSICAL LAYER: NETWORK RESILIENCE</strong></p><p>Network Reliability: Measured by the probability that a network performs satisfactorily over a period of time.</p><p>Measures include:</p><ol><li>Mean Time Between Failures (MTBF)</li><li>Mean Time To Failure (MTTF)</li><li>Mean Time to Repair (MTTR)</li></ol><p><strong>MTBF = MTTF + MTTR</strong></p><p><strong>Link Failure Probability</strong>: Percentage of time link is down</p><p><strong>Link Availability</strong>: Percentage of time link is up</p><p><strong>Network Fault Tolerance: </strong>Probability that the network remains connected.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=dbf7cd5626c0" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Network Science ExClass [1]]]></title>
            <link>https://medium.com/@hjchew96/network-science-exclass-1-6eb402fc6e88?source=rss-7956c6ed626d------2</link>
            <guid isPermaLink="false">https://medium.com/p/6eb402fc6e88</guid>
            <dc:creator><![CDATA[Hong Jun Chew]]></dc:creator>
            <pubDate>Tue, 11 Feb 2020 08:55:31 GMT</pubDate>
            <atom:updated>2020-02-11T08:55:31.027Z</atom:updated>
            <content:encoded><![CDATA[<h4>Example Class 1</h4><ol><li>Directed and Undirected network<br>for an undirected network, adjacency matrics should be asymmetric<br>Hence, the network is directed</li><li>In-degree, Out-degree<br>For directed Graphs,<br>Number of links pointing in<br>Number of links pointing out</li><li>How many weakly connected components</li></ol><blockquote>a weakly connected component (WCC) is a subgraph of the original graph where <strong>all vertices are connected to each other by some path, ignoring the direction of edges(subgraph is fully connected)</strong></blockquote><p>4. Compute clustering coefficient Cᵢ<br>Cᵢ = 2eᵢ/kᵢ(kᵢ-1) or,<br>Cᵢ = 2(number of triangle)/degree(degree-1)</p><p>5. Drawbacks of Metcalfe’s Law</p><p>Metcalfe states that the value of a particular network is given by the square of the number of nodes and assumes each link to be equally weighted. Which is not true in real world networks</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6eb402fc6e88" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Network Science ExClass [2]]]></title>
            <link>https://medium.com/@hjchew96/network-science-exclass-2-61b1a1811910?source=rss-7956c6ed626d------2</link>
            <guid isPermaLink="false">https://medium.com/p/61b1a1811910</guid>
            <dc:creator><![CDATA[Hong Jun Chew]]></dc:creator>
            <pubDate>Tue, 11 Feb 2020 04:22:44 GMT</pubDate>
            <atom:updated>2020-02-11T04:22:44.381Z</atom:updated>
            <content:encoded><![CDATA[<h4>Example class 2</h4><p>Consider a random network G(N,p)<br>N nodes, link probability p<br>N = 10001, p = 10^-2</p><ol><li>Expected number of links <strong>&lt;L&gt; = p(N)(N-1)/2</strong> = (10^-2)(10001)(10000) = 500050</li><li>Average Degree <strong>&lt;k&gt; = p(N-1) = 2&lt;L&gt;/N</strong> = (10^-2)(10000) = 100</li><li>There is a <strong>giant component</strong> if average degree<strong>&lt;k&gt; exceeds 1</strong></li><li>lnN = 9.21, given that average degree <strong>&lt;k&gt; &gt; 9.21</strong>. Therefore this is a connected regime.</li><li>critical point &lt;k&gt;=1, <strong>&lt;k&gt; = p(N-1)</strong>,<strong> p = &lt;k&gt;/(N-1)</strong> = 1/(N-1) = 1/10000 = 10^-4</li><li>avg distance,<strong> &lt;d&gt; = logN/log&lt;k&gt;</strong> = log10001/log100 = 2</li><li><strong>p(k) = e^-&lt;k&gt; ( &lt;k&gt;^k /k! )</strong></li><li><strong>&lt;L&gt; = p(n*[n-1]/2)<br>&lt;k&gt; = p(n-1)</strong></li><li>Q5… im not sure :(</li></ol><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=61b1a1811910" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Network Science Notes [4]]]></title>
            <link>https://medium.com/@hjchew96/network-science-notes-4-e122e982cf13?source=rss-7956c6ed626d------2</link>
            <guid isPermaLink="false">https://medium.com/p/e122e982cf13</guid>
            <dc:creator><![CDATA[Hong Jun Chew]]></dc:creator>
            <pubDate>Mon, 10 Feb 2020 04:01:44 GMT</pubDate>
            <atom:updated>2020-02-10T04:53:59.063Z</atom:updated>
            <content:encoded><![CDATA[<h3><strong>Scale-Free Property</strong></h3><p>Scale-Free Networks are those that follow the <strong><em>power law</em></strong><br>pₖ = CK ⁻γ</p><p>A network is called <strong>scale</strong>-<strong>free</strong> if the characteristics of the network are independent of the size of the network, i.e. the number of nodes</p><p><strong>Questions</strong></p><ol><li>What’s the difference between <em>power-law</em> and <em>Poisson distribution</em>?</li><li>What does <em>scale-free</em> mean?</li><li>Is scale-free property of <em>universal networks</em>?</li><li>How do <em>hubs</em> affect the <em>small-world property</em>?</li></ol><h4>1. differences between power-law and Poisson distribution?</h4><p>The biggest difference between power-law and Poisson distribution is in the tail of the distribution. In the high-k regime(probability of a node with a high degree), in a Poisson distribution, the probability is significantly underestimated compared to the power-law.</p><p>As seen from the diagram below a Poisson distribution(Bell curve), significantly underestimates the probability of nodes with a high number of k.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zOkfv5lSkUtSk1Da76VzZQ.png" /><figcaption><a href="https://slideplayer.com/slide/14359722/">https://slideplayer.com/slide/14359722/</a></figcaption></figure><p>kₘₐₓ = kₘᵢₙN¹/ʸ-¹</p><p>expected maximum degree of hubs</p><p><strong>“Scale-Free”</strong></p><p>significance of lower moments</p><p>n=1: avg degree&lt;k&gt;<br>n=2: variance &lt;k²&gt;-&lt;k&gt;²<br>n=3: skewness</p><p>if n ≲ γ -1: all moments will be finite<br>if n ≳ γ -1: all moments will diverge</p><p>_________________________________</p><p>for Free scale networks, n is between 2 and 3</p><p>The average degree &lt;k&gt; acts as the scale of a random network</p><p>This is hard I don’t get it</p><p><strong>Is scale-free property of <em>universal networks</em></strong></p><p>scale-free is a property of</p><p>Nodes: computers &amp; routers<br>Links: physical lines</p><p>Nodes: Scientist<br>Links: a joint publication</p><p>However, networks in materials and neural networks are generally not free scale.</p><p><strong>How do <em>hubs</em> affect the <em>small-world property</em></strong></p><p>Average distance follows the small world of random networks<br>Hubs are not sufficiently large enough</p><p>… not sure what to write</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e122e982cf13" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Network Science Notes [3]]]></title>
            <link>https://medium.com/@hjchew96/network-science-notes-3-d8bb9ac6fb32?source=rss-7956c6ed626d------2</link>
            <guid isPermaLink="false">https://medium.com/p/d8bb9ac6fb32</guid>
            <dc:creator><![CDATA[Hong Jun Chew]]></dc:creator>
            <pubDate>Mon, 10 Feb 2020 03:56:53 GMT</pubDate>
            <atom:updated>2020-02-10T03:56:53.444Z</atom:updated>
            <content:encoded><![CDATA[<h4><strong>Are Real Networks like random graphs</strong></h4><p><strong>important concepts:</strong></p><p>l: Length<br>N: Number of Nodes<br>k: degree<br>&lt;k&gt;: Average degree<br>Average Path Length: &lt;lᵣₐₙ&gt; = log N / log&lt;k&gt;<br>Clustering Coefficient: Cᵢ = 2&lt;Lᵢ&gt;/Kᵢ(Kᵢ-1) = &lt;k&gt;/N<br>Degree Distribution: = eᵃᵛᵍ⁻ᵏ * &lt;k&gt;ᵏ/k!</p><p><strong>Ans:</strong></p><ol><li>Real networks have short distances like a random graph.</li><li>Cᵣₐₙ under estimates clustering coefficients by orders of magnitude.</li></ol><p>Random Networks: Nodes N, have a probability P that is connected.<br>Random networks are able to explain small-world phenomenon¹<br>However, they are not able to explain the high degree of cliquishness²</p><p>All in all, we can conclude that Real networks are not completely random, Even something like proteins, social media</p><h4><strong>Watts-Strogatz Model</strong></h4><p>Purpose: Overcome the problem of randomness and cliquishness.<br>Model: <br>1. The starting point has a large clustering coefficient.<br>2. Allowing links to be drop-out at a fixed probability(<em>p</em>).<br>3. Effective interpolates between <strong>regular lattice(<em>p</em> → 0)</strong> and a <strong>random network (<em>p →1)</em></strong><em>.</em></p><p>Observations:<br>Even at small values of <em>P, </em>the network still exhibits <strong>regularity </strong>with a <strong>high clustering coefficient. </strong>This small value of <em>p, </em>creates short cuts between distant nodes, reducing the <strong>overall distance.</strong></p><p><strong>As P increases from 0(no drop) to 1(complete randomness) we observe:<br>↑ </strong>a fast decrease in average distance<strong><br>↓ </strong>a slow decrease in clustering coefficient</p><p>This shows that Watts-Strogatz Models was better at modeling the real world than a completely random network.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d8bb9ac6fb32" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Network Science Notes [2]]]></title>
            <link>https://medium.com/@hjchew96/network-science-notes-2-cb49db172c4d?source=rss-7956c6ed626d------2</link>
            <guid isPermaLink="false">https://medium.com/p/cb49db172c4d</guid>
            <dc:creator><![CDATA[Hong Jun Chew]]></dc:creator>
            <pubDate>Mon, 10 Feb 2020 03:55:52 GMT</pubDate>
            <atom:updated>2020-02-19T03:51:34.796Z</atom:updated>
            <content:encoded><![CDATA[<h3>Network Science[2]</h3><p>Questions of the day:<br>1. Can a random network model explain the evolution of networks? <br>2. Can it model the small world phenomenon?<br>3. Can it capture the local clustering of real networks?</p><p><strong>N: </strong>Number of Nodes<strong><br>Ng: </strong>Number of nodes in a giant component<strong><br>&lt;K&gt;: </strong>Average Degree</p><p><strong>Quantifying Evolution<br></strong>Investigate the relationship between Ng and &lt;k&gt;<br>Giant component grows from Ng =1 → Ng = N <br>when &lt;k&gt; increase from 0→N-1.</p><p>For a <em>Giant Component</em> to exist fall nodes have to <em>have at least 1 link</em> thus, the average degree should be at least &lt;k&gt; = 1</p><p>I: Subcritical → II: Critical → III: Supercritical → IV: Connected<br>&lt;k&gt; &lt; 1 — — — &lt;k&gt; = 1 — — — &lt;k&gt; &gt; 1 — — — &lt;k&gt; &gt; lnN</p><p>Average Degree <strong>&lt;k&gt; </strong>: P(N-1)<br>For subcritical P(N-1) &lt; 1 == p = 1/N</p><p>The <em>Largest possible cluster</em> is a tree, size ~ <em>lnN</em></p><p><strong>Real Networks</strong></p><p>Once &lt;k&gt; &gt; 1, a giant component(GC) emerges<br>For &lt;k&gt; &gt; 1, GC looks like a network<br>&lt;k&gt; = lnN, all component observed</p><p><strong>Weakly Connected Component (WCC)</strong></p><ul><li>For a directed graph, a weakly connected component (WCC) is a subgraph of the original graph where <strong>all vertices are connected to each other by some path, ignoring the direction of edges(subgraph is fully connected)</strong></li><li>In case of an undirected graph, a weakly connected component is also a strongly connected component</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/757/1*OIncqjZ7PPfQ20s6glAIwQ.png" /><figcaption><a href="https://www.geeksforgeeks.org/strongly-connected-components/">https://www.geeksforgeeks.org/strongly-connected-components/</a></figcaption></figure><blockquote>A directed graph is strongly connected if there is a path between all pairs of zvertices. A strongly connected component (<strong>SCC</strong>) of a directed graph</blockquote><p><strong>Clustering Coefficient</strong></p><p>The clustering coefficient of an undirected graph is a measure of the number of triangles in a graph</p><p>Cᵢ = number of triangles connected to node i/number of triples centered around node i</p><p>CC(v): <br>v: node:<br>kᵥ = degree<br>nᵥ = number of links</p><p>kᵥ = 4, nᵥ = 1<br>cc(v) = 2 nᵥ/kᵥ(kᵥ-1)</p><p><a href="https://en.wikipedia.org/wiki/Metcalfe%27s_law"><strong>Metcalfe’s Law</strong></a></p><blockquote>Metcalfe’s Law states the effect of a telecommunications network is proportional to the square of the number of connected user of the system (n²)</blockquote><p>Metcalfe’s Law is based off on how for a random network with <strong><em>n</em></strong> nodes there can be expressed as n(n-1)/2, the number of triangles. which is proportional to n²</p><p><strong>The Party</strong></p><blockquote>If each person meets at least one other guest, the information will reach most quests</blockquote><p>Once the average degree &lt;k&gt; = 1, a giant component, GC emerges containing some of all nodes</p><p>&lt;k&gt; &gt; 1, GC becocomes into a recognizable network</p><p>&lt;k&gt; &gt; lnN, all components are connected to GC</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cb49db172c4d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Network Science Notes [1]]]></title>
            <link>https://medium.com/@hjchew96/network-science-notes-1-adf23e4d9cab?source=rss-7956c6ed626d------2</link>
            <guid isPermaLink="false">https://medium.com/p/adf23e4d9cab</guid>
            <dc:creator><![CDATA[Hong Jun Chew]]></dc:creator>
            <pubDate>Tue, 04 Feb 2020 04:36:01 GMT</pubDate>
            <atom:updated>2020-02-10T03:09:56.207Z</atom:updated>
            <content:encoded><![CDATA[<h3>Network Science [1]</h3><p>How can we build models that accurately reproduce the properties of real networks observed in real systems?</p><p><strong>Random Network</strong>: Graph of <strong>N</strong> nodes where each pair have the same probability <strong>P</strong> of connection.</p><p>G(N, L) Model: <strong>N</strong> number of nodes, with <strong>L</strong> number of fixed links.<br>G(N, P) Model: each <strong>N</strong> labeled node are connected with a probability <strong>P<br></strong>The degree of a vertice, U: is the number of edges/links connected to it.</p><p><strong>Directed Vs Undirected Network</strong></p><ul><li>Directed Network(Digraph), edges are directed from one vertex to another</li><li>Undirected Network, edges are not directed from one vertex to another</li></ul><p>Undirected Networks will, therefore, be symmetrical across y=-x in an adjacency matrix</p><p><strong>Network Distance</strong></p><ul><li>Distance between vertices u, v: Number of edges on the shortest path from u to v</li><li>Average Distance &lt;d&gt; for a connected graph:<br>&lt;d&gt; = [1/2Lₘₐₓ]∑dᵢⱼ<br>&lt;d&gt; = log N/log&lt;k&gt;</li></ul><p><strong>Network Diameter</strong></p><ul><li>Diameter<br>worst case diameter: Largest Distance between a pair of coords.<br>Average case diameter: Average distance between coords.</li></ul><p><strong>Random Network Construction</strong></p><p>For a network with N nodes → for each pair randomly generate a link with probability P → Repeat for N(N-1)/2.</p><p>P = 1/6<br>N = 10</p><p>&lt;k&gt; ~1.5<br>&lt;k&gt;, Average Degree = E*2/N</p><p>Equations to know:<br>1. The probability that the random network has exactly L links (P*(1-P))<br>2. Expected Number of Links (P * number of nodes)<br>3. Average Degree ( 2*Expected Number of Links / N-1 )<br>4. Standard Deviation (p*(1-P)*number of links)</p><p><strong>Degree Distribution</strong></p><p>distribution of link per node</p><p>P(k) = (number of permutations to select K links) *(probability of K links) *(probability of others)</p><p>Standard deviation /Average degree ~ 1/(N-1)²</p><p><strong>Clustering Coefficient</strong></p><p>Kᵥ = node’s degree<br>Nᵥ = number of links between neighbor, i.e number of triangles <br>cc(v) = 2Nᵥ/kᵥ(kᵥ-1)</p><p><strong>Most Networks are Spare</strong></p><p>&lt;k&gt; &lt;&lt; N<br>Degree Distribution can be approximated by <strong>Poisson Distribution</strong></p><p>References:<br><a href="http://barabasi.com/f/624.pdf">http://barabasi.com/f/624.pdf</a></p><p>End</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=adf23e4d9cab" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>