https://wiki-planet.win/api.php?action=feedcontributions&feedformat=atom&user=Carinengbb 2026-06-01T21:51:35Z User contributions MediaWiki 1.42.3 https://wiki-planet.win/index.php?title=Questions_for_Event_Agencies_in_Malaysia_Before_Reservoir_Computing_Forums_to_Stay_Organized&diff=2007247 2026-05-28T17:38:09Z

<p>Carinengbb: Created page with &quot;&lt;html&gt;&lt;p class=&quot;ds-markdown-paragraph&quot; &gt; Echo state networks are not conventional deep learning. Standard neural networks train all connections. Liquid state machines only adjust the final connections. The hidden pool is unchanging and arbitrary. This leads to accelerated training and demands smaller datasets.&lt;/p&gt;&lt;p class=&quot;ds-markdown-paragraph&quot; &gt; A reservoir computing forum is not a standard AI conference. It must address reservoir dynamics, spectral radius, leakage r...&quot;</p> <hr /> <div>&lt;html&gt;&lt;p class=&quot;ds-markdown-paragraph&quot; &gt; Echo state networks are not conventional deep learning. Standard neural networks train all connections. Liquid state machines only adjust the final connections. The hidden pool is unchanging and arbitrary. This leads to accelerated training and demands smaller datasets.&lt;/p&gt;&lt;p class=&quot;ds-markdown-paragraph&quot; &gt; A reservoir computing forum is not a standard AI conference. It must address reservoir dynamics, spectral radius, leakage rate, and readout training (ridge regression).&lt;/p&gt;&lt;p class=&quot;ds-markdown-paragraph&quot; &gt; Businesses questioning coordinators in Klang Valley for reservoir computing forums|for echo state network summits|for liquid state machine gatherings need technical questions|require specific inquiries|must ask targeted queries.&lt;/p&gt;&lt;h2&gt; Why &quot;It Works&quot; Is Not Enough&lt;/h2&gt;&lt;p&gt; &lt;img src=&quot;https://i.ytimg.com/vi/RI35E5ewBuI/hq720.jpg&quot; style=&quot;max-width:500px;height:auto;&quot; &gt;&lt;/img&gt;&lt;/p&gt;&lt;p class=&quot;ds-markdown-paragraph&quot; &gt; Some coordinators might showcase echo state networks without showing the echo state property. The fading memory guarantees that the reservoir's state depends on recent inputs, not initial conditions.&lt;/p&gt;&lt;p class=&quot;ds-markdown-paragraph&quot; &gt; A representative from once told me: “A vendor claimed a reservoir computing demo. They ran a script. It produced outputs. I asked 'how do you know the echo state property holds?' They looked confused. 'What is echo state?' they asked. They were using random weights but had no idea if the reservoir had memory. The demo was useless. Now we ask every agency: 'Do you verify the echo state property before your demo?'”&lt;/p&gt;&lt;p&gt; &lt;iframe src=&quot;https://www.youtube.com/embed/5O6U4a6Ej_M&quot; width=&quot;560&quot; height=&quot;315&quot; style=&quot;border: none;&quot; allowfullscreen=&quot;&quot; &gt;&lt;/iframe&gt;&lt;/p&gt;&lt;p class=&quot;ds-markdown-paragraph&quot; &gt; Pose these questions to coordinators: Do you confirm the short-term retention of the hidden layer. What are the scaling factors of your hidden connections, and how were they determined.&lt;/p&gt;&lt;h2&gt; Why &quot;We Use a Dense Layer&quot; Is Not Reservoir Computing&lt;/h2&gt;&lt;p class=&quot;ds-markdown-paragraph&quot; &gt; Some vendors claim reservoir computing but train the reservoir. This is not reservoir computing. Only the output weights should be adjusted.&lt;/p&gt;&lt;p class=&quot;ds-markdown-paragraph&quot; &gt; Talk through with your coordinator: Does your showcase learn only the readout, or do you also modify internal connections. What optimization technique do you employ for output weights (ridge regression, LASSO, or elastic net).&lt;/p&gt;&lt;p class=&quot;ds-markdown-paragraph&quot; &gt; A reservoir computing scientist from KL wrote: “I attended a 'reservoir computing' event where the presenter trained the reservoir using backpropagation. I asked 'why are you training the reservoir?' He said 'it improves performance.' I said 'then it is not reservoir computing. Reservoir computing means fixed reservoir, trained readout. You are just doing a small recurrent network.' He had no answer. The event was misleading.”&lt;/p&gt;&lt;h2&gt; The Temporal Task: Showcasing Memory&lt;/h2&gt;&lt;p class=&quot;ds-markdown-paragraph&quot; &gt; Reservoir computing's strength is time-dependent information, future value forecasting, and ordered input handling.&lt;/p&gt;&lt;p&gt; &lt;img src=&quot;https://i.ytimg.com/vi/GSmKwiUc2mo/hq720.jpg&quot; style=&quot;max-width:500px;height:auto;&quot; &gt;&lt;/img&gt;&lt;/p&gt;&lt;p class=&quot;ds-markdown-paragraph&quot; &gt; A non-temporal task (like image recognition) does not highlight echo state networks.&lt;/p&gt;&lt;p class=&quot;ds-markdown-paragraph&quot; &gt; Pose these questions to coordinators: What temporal task will you demonstrate (e.g., NARMA series prediction, Mackey-Glass time series, or sine wave generation).&lt;/p&gt;&lt;h2&gt; The Hyperparameter Discussion: Spectral Radius, Leakage Rate, Input Scaling&lt;/h2&gt;&lt;p&gt; &lt;img src=&quot;https://i.ytimg.com/vi/xZKse0mEpfg/hq720.jpg&quot; style=&quot;max-width:500px;height:auto;&quot; &gt;&lt;/img&gt;&lt;/p&gt;&lt;p class=&quot;ds-markdown-paragraph&quot; &gt; Reservoir computing has critical hyperparameters. Spectral radius (should be slightly less than 1). Fading speed (for analog-time pools). Input scaling (connects input size to reservoir dynamics).&lt;/p&gt;&lt;p class=&quot;ds-markdown-paragraph&quot; &gt; &lt;a href=&quot;https://www.balaken.info/user/cromlirnxy&quot;&gt;event planner malaysia&lt;/a&gt; recommends an interactive setting demonstration showing how results shift with different adjustments.&lt;/p&gt;&lt;/html&gt;</div>

Carinengbb